Method for performing communication in wireless communication system and device therefor

Abstract

Disclosed are a method for performing communication in a wireless communication system according to various embodiments and a device therefor. Disclosed are a method and a device therefor, the method comprising: transmitting, to a base station, a signal for data transmission and sensing; receiving a reflected signal of the signal so as to acquire first radar data for a first frequency band; and receiving, from the base station, second radar data including synthetic radar data generated on the basis of the first radar data.

Description

Method for performing communication in a wireless communication system and device for the same

[0001] This relates to a method for a terminal to perform communication in a wireless communication system and a device for doing so.

[0002] Various devices and technologies, such as machine-to-machine (M2M) communication, machine type communication (MTC), and devices requiring high data transmission rates like smartphones and tablet PCs (Personal Computers), are emerging and becoming widespread. Consequently, the amount of data required to be processed in cellular networks is increasing very rapidly. To satisfy this rapidly increasing demand for data processing, technologies such as carrier aggregation and cognitive radio are being developed to efficiently utilize more frequency bands, while technologies such as multi-antenna technology and multi-BS cooperation are being developed to increase the data capacity transmitted within a limited frequency range.

[0003] As more communication devices require greater communication capacity, the need for enhanced mobile broadband (eMBB) communication is emerging compared to legacy radio access technology (RAT). In addition, massive machine type communication (mMTC), which connects multiple devices and objects to provide various services anytime and anywhere, is one of the key issues to consider in next-generation communication.

[0004] In addition, discussions are underway regarding communication systems to be designed with user equipment (UE) in mind, which is sensitive to reliability and latency. The introduction of next-generation wireless access technologies is being discussed with consideration of eMBB communication, mMTC, and ultra-reliable and low-latency communication (URLLC).

[0005] The technical objective to be achieved is to provide a method for effectively applying artificial neural networks to wireless communication systems to provide various types of communication services.

[0006] The technical problems that this specification aims to solve are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art related to this specification from the detailed description below.

[0007] A method by a terminal according to one aspect comprises the steps of: transmitting a signal for data transmission and sensing to a base station; receiving a reflected signal of the signal to acquire first radar data for a first frequency band and transmitting the first radar data to the base station; and receiving second radar data including synthetic radar data generated based on the first radar data from the base station, wherein the synthetic radar data may be radar data synthesized based on the first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

[0008] Alternatively, the second frequency band is characterized by having a starting frequency that is higher than the end frequency of the first frequency band by a preset frequency, and having a frequency bandwidth wider than the bandwidth of the first frequency band.

[0009] Alternatively, the first radar data is characterized by being synthesized into the synthetic radar data through sequential processing by a first model based on VAE (Variational Autoencoder), a second model based on DM (Diffusion Model), and a third model based on GAN (Generative Adversarial Networks).

[0010] Alternatively, the above VAE-based first model is characterized as a model that outputs radar data that compensates for missing values ​​for a bandwidth wider than the bandwidth of the first frequency band based on the above first radar data.

[0011] Alternatively, the DM-based second model is characterized as a model that outputs refined radar data by removing noise from the output radar data of the VAE-based first model through input forward diffusion and reverse diffusion.

[0012] Alternatively, the GAN-based third model is a model that outputs synthesized radar data synthesized to have frequency characteristics for a second frequency band higher than the first frequency band based on the first radar data, wherein the second frequency band has a starting frequency that is higher by a preset frequency than the end frequency of the first frequency band and has a frequency bandwidth wider than the bandwidth of the first frequency band.

[0013] Alternatively, the above VAE-based first model outputs supplementary radar data that supplements missing values ​​for a bandwidth wider than the bandwidth of the first frequency band based on the first radar data, the above DM-based second model receives the supplementary radar data and outputs refined radar data that removes noise from the supplementary radar data through forward diffusion and reverse diffusion, and the above GAN-based third model receives the refined radar data and outputs the synthesized radar data synthesized to mimic the frequency characteristics of a second frequency band higher than the first frequency band based on the refined radar data.

[0014] Alternatively, the terminal is characterized by training a sensing operation related to ISAC (Integrated Sensing and Communication) or a model based on an artificial neural network related to ISAC based on the second radar data.

[0015] Depending on other aspects, a non-transient computer-readable storage medium may be provided that records instructions for performing the method by the terminal described above.

[0016] Depending on another aspect, a terminal that performs the method described above may be provided.

[0017] A method by a base station according to another aspect comprises: receiving first radar data for a first frequency band based on a reflected signal of a signal for data transmission and sensing from a terminal; generating synthetic radar data related to the first radar data through sequential processing of a first model based on a Variational Autoencoder (VAE), a second model based on a Diffusion Model (DM), and a third model based on Generative Adversarial Networks (GAN); and transmitting second radar data including the synthetic radar data to the terminal, wherein the synthetic radar data may be radar data synthesized based on the first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

[0018] Depending on other aspects of the work, a base station for performing the method described above may be provided.

[0019] The above-mentioned problem-solving methods are merely some of the examples in this specification, and various examples reflecting the technical features of this specification can be derived and understood by a person skilled in the art based on the following detailed description.

[0020] According to some implementations of the present specification, an artificial neural network model for providing various types of communication services can be effectively applied to a wireless communication system.

[0021] The effects according to this specification are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art related to this specification from the detailed description below.

[0022] The accompanying drawings, which are included as part of the detailed description to aid in understanding the implementations of this specification, provide examples of the implementations of this specification and describe the implementations of this specification together with the detailed description:

[0023] FIG. 1 illustrates an example of a communication system 1 to which the implementations of the present specification are applied;

[0024] FIG. 2 is a block diagram illustrating examples of communication devices capable of performing the method according to the present specification;

[0025] FIG. 3 illustrates another example of a wireless device capable of performing the implementation(s) of the present specification;

[0026] FIG. 4 illustrates an example of a frame structure available in a 3rd generation partnership project (3GPP)-based wireless communication system;

[0027] FIG. 5 illustrates physical channels used in a 3rd generation partnership project (3GPP)-based communication system, which is an example of a wireless communication system, and the signal transmission / reception process using them;

[0028] FIG. 6 illustrates an arbitrary connection process that may be applied to the implementation(s) of the present specification;

[0029] FIG. 7 illustrates a perceptron structure used in an artificial neural network;

[0030] FIG. 8 illustrates a multilayer perceptron structure;

[0031] Figure 9 illustrates a convolutional neural network (CNN) structure;

[0032] FIG. 10 illustrates a filter operation in a CNN;

[0033] Figure 11 illustrates the concept of step-to-step learning with backpropagation applied;

[0034] Figure 12 is an example of a method for calculating gradients in a neural network;

[0035] Figure 13 illustrates the structure of a GAN.

[0036] Figure 14 is a diagram illustrating a VAE model.

[0037] FIG. 15 illustrates examples of wireless sensing modes supported by ISAC.

[0038] Figures 16 and 17 show an example of how ISAC is applied to a 3GPP wireless communication system.

[0039] FIG. 18 is a diagram illustrating how a device performs ISAC using an AI model.

[0040] Figure 19 is a diagram illustrating how a base station and a terminal process radar data.

[0041] FIG. 20 is a diagram illustrating a method for a terminal to receive second radar data generated based on first radar data from a base station.

[0042] FIG. 21 is a diagram illustrating a method for a base station to acquire second reflection data generated based on first radar data received from a terminal.

[0043] Hereinafter, implementations according to the present specification will be described in detail with reference to the accompanying drawings. The detailed description disclosed below, together with the accompanying drawings, is intended to describe exemplary implementations of the present specification and is not intended to represent the only form in which the present specification may be practiced. The following detailed description includes specific details to provide a complete understanding of the present specification. However, those skilled in the art will know that the present specification may be practiced without such specific details.

[0044] In some cases, to avoid ambiguity of the concepts of this specification, known structures and devices may be omitted or illustrated in the form of block diagrams focusing on the core functions of each structure and device. Additionally, throughout this specification, the same reference numerals are used to describe identical components.

[0045] The techniques, devices, and systems described below can be applied to various wireless multiple access systems. Examples of multiple access systems include CDMA (code division multiple access) systems, FDMA (frequency division multiple access) systems, TDMA (time division multiple access) systems, OFDMA (orthogonal frequency division multiple access) systems, SC-FDMA (single carrier frequency division multiple access) systems, and MC-FDMA (multi carrier frequency division multiple access) systems. CDMA can be implemented in wireless technologies such as UTRA (Universal Terrestrial Radio Access) or CDMA2000. TDMA can be implemented in wireless technologies such as GSM (Global System for Mobile communication), GPRS (General Packet Radio Service), and EDGE (Enhanced Data Rates for GSM Evolution) (i.e., GERAN). OFDMA can be implemented in wireless technologies such as IEEE (Institute of Electrical and Electronics Engineers) 802.11 (WiFi), IEEE 802.16 (WiMAX), IEEE 802-20, and E-UTRA (evolved-UTRA). UTRA is part of UMTS (Universal Mobile Telecommunication System), and 3GPP (3rd Generation Partnership Project) LTE (Long Term Evolution) is part of E-UMTS that utilizes E-UTRA.3GPP LTE adopts OFDMA for the downlink (DL) and SC-FDMA for the uplink (UL). LTE-A (LTE-advanced) is an evolved form of 3GPP LTE.

[0046] For convenience of explanation, the following description assumes that this specification applies to 3GPP-based communication systems, e.g., LTE and NR. However, the technical features of this specification are not limited thereto. For example, even though the following detailed description is based on a mobile communication system corresponding to a 3GPP LTE / NR system, it is applicable to any other mobile communication system except for matters specific to 3GPP LTE / NR.

[0047] For terms and technologies used in this specification that are not specifically described, reference may be made to 3GPP-based standard documents, for example, 3GPP TS 36.211, 3GPP TS 36.212, 3GPP TS 36.213, 3GPP TS 36.321, 3GPP TS 36.300 and 3GPP TS 36.331, 3GPP TS 37.213, 3GPP TS 38.211, 3GPP TS 38.212, 3GPP TS 38.213, 3GPP TS 38.214, 3GPP TS 38.300, 3GPP TS 38.304, 3GPP TS 38.331, etc.

[0048] In the examples of this specification set forth below, the expression that the device "assumes" may mean that the entity transmitting the channel transmits the channel in accordance with said "assume." It may mean that the entity receiving the channel receives or decodes the channel in a form that conforms to said "assume," under the premise that the channel was transmitted in accordance with said "assume."

[0049] In this specification, ' / ' may mean 'and / or'. For example, cell DTX / DRX may mean cell DTX and / or cell DRX.

[0050] In this specification, a UE may be fixed or mobile and includes various devices that communicate with a base station (BS) to transmit and / or receive user data and / or various control information. A UE may be referred to as Terminal Equipment, Mobile Station (MS), Mobile Terminal (MT), User Terminal (UT), Subscribe Station (SS), wireless device, Personal Digital Assistant (PDA), wireless modem, handheld device, etc. Additionally, in this specification, a BS generally refers to a fixed station that communicates with a UE and / or other BSs, and exchanges various data and control information by communicating with a UE and other BSs. A BS may be referred to by other terms such as Advanced Base Station (ABS), Node-B (NB), eNB (evolved-NodeB), Base Transceiver System (BTS), Access Point, Processing Server (PS), etc. In particular, BSs of UTRAN are called Node-Bs, BSs of E-UTRAN are called eNBs, and BSs of new radio access technology networks are called gNBs. For convenience of explanation, BSs will be collectively referred to as BSs regardless of the type or version of the communication technology.

[0051] In this specification, the term "node" refers to a fixed point capable of transmitting / receiving wireless signals by communicating with a UE. Various types of BSs may be used as nodes regardless of their designation. For example, a BS, NB, eNB, pico-cell eNB (PeNB), home eNB (HeNB), relay, repeater, etc., may serve as a node. Additionally, a node may not be a BS. For example, it may be a radio remote head (RRH) or a radio remote unit (RRU). RRHs, RRUs, etc. generally have a power level lower than that of a BS. Since an RRH or RRU (or RRH / RRU) is generally connected to a BS via a dedicated line such as an optical cable, cooperative communication between an RRH / RRU and a BS can be performed more smoothly compared to cooperative communication between BSs connected via a wireless line. At least one antenna is installed at a node. This antenna may refer to a physical antenna, an antenna port, a virtual antenna, or an antenna group. Nodes are also referred to as points.

[0052] In this specification, the term "cell" refers to a specific geographical area where one or more nodes provide communication services. Accordingly, in this specification, communicating with a specific cell may mean communicating with a BS or node that provides communication services to said specific cell. Furthermore, the downlink / uplink signal of a specific cell refers to a downlink / uplink signal from to or to the BS or node that provides communication services to said specific cell. A cell that provides uplink / downlink communication services to a UE is specifically referred to as a serving cell. Additionally, the channel state / quality of a specific cell refers to the channel state / quality of a channel or communication link formed between the BS or node providing communication services to said specific cell and the UE. In a 3GPP-based communication system, a UE can measure the downlink channel state from a specific node using the CRS(s) transmitted by the antenna port(s) of the specific node over the CRS (Cell-specific Reference Signal) resource assigned to the specific node and / or the CSI-RS(s) transmitted over the CSI-RS (Channel State Information Reference Signal) resource.

[0053] Meanwhile, 3GPP-based communication systems use the concept of a cell to manage wireless resources, and a cell associated with wireless resources is distinguished from a cell in a geographical area.

[0054] A “cell” of a geographical area can be understood as the coverage over which a node can provide services using a carrier wave, and a “cell” of a wireless resource is associated with the bandwidth (BW), which is the frequency range configured by said carrier wave. Since downlink coverage, which is the range over which a node can transmit a valid signal, and uplink coverage, which is the range over which a valid signal can be received from a UE, depend on the carrier wave carrying the signal, the coverage of a node is also associated with the coverage of the “cell” of the wireless resource used by said node. Therefore, the term “cell” can be used to refer sometimes to the coverage of a service by a node, sometimes to a wireless resource, and sometimes to the range over which a signal using said wireless resource can reach with effective strength.

[0055] Meanwhile, 3GPP communication standards use the concept of a cell to manage radio resources. A "cell" associated with radio resources is defined as a combination of downlink resources (DL resources) and uplink resources (UL resources), that is, a combination of a DL component carrier (CC) and a UL CC. A cell can be configured as a DL resource alone or as a combination of a DL resource and a UL resource. Where carrier aggregation is supported, the linkage between the carrier frequency of a DL resource (or DL ​​CC) and the carrier frequency of a UL resource (or UL CC) can be indicated by system information. For example, the combination of DL resources and UL resources can be indicated by a System Information Block Type 2 (SIB2) linkage. Here, the carrier frequency may be equal to or different from the center frequency of each cell or CC. When Carrier Aggregation (CA) is established, the UE has only one Radio Resource Control (RRC) connection with the network. One serving cell provides Non-Access Stratum (NAS) mobility information during RRC establishment / re-establishment / handover, and one serving cell provides security input during RRC re-establishment / handover. This cell is called a primary cell (Pcell). A Pcell is a cell operating on the primary frequency where the UE performs the initial connection establishment procedure or initiates the connection re-establishment procedure.Depending on the UE capability, secondary cells (Scells) can be configured to form a set of serving cells together with Pcells. Scells can be configured after a Radio Resource Control (RRC) connection is established and are cells that provide additional radio resources in addition to the resources of special cells (SpCells). The carrier corresponding to a Pcell in the downlink is called the Downlink Primary CC (DL PCC), and the carrier corresponding to a Pcell in the uplink is called the UL Primary CC (UL PCC). The carrier corresponding to an Scell ​​in the downlink is called the DL Secondary CC (DL SCC), and the carrier corresponding to the Scell ​​in the uplink is called the UL Secondary CC (UL SCC).

[0056] In a wireless communication system, the UE receives information from the BS via the downlink (DL) and transmits information to the BS via the uplink (UL). The information transmitted and / or received by the BS and the UE includes data and various control information, and various physical channels exist depending on the type and purpose of the information they transmit and / or receive.

[0057] 3GPP-based communication standards define downlink physical channels corresponding to resource elements that carry information originating from upper layers, and downlink physical signals corresponding to resource elements used by the physical layer but not carrying information originating from upper layers. For example, the physical downlink shared channel (PDSCH), physical broadcast channel (PBCH), and physical downlink control channel (PDCCH) are defined as downlink physical channels, while the reference signal and synchronization signal are defined as downlink physical signals. The reference signal (RS), also referred to as a pilot, refers to a signal of a specific, predefined waveform known to both the BS and the UE. For example, the demodulation reference signal (DMRS), channel state information RS (CSI-RS), and positioning reference signal (PRS) are defined as downlink reference signals. 3GPP-based communication standards define uplink physical channels corresponding to resource elements that carry information originating from upper layers, and uplink physical signals corresponding to resource elements that are used by the physical layer but do not carry information originating from upper layers.For example, the physical uplink shared channel (PUSCH), physical uplink control channel (PUCCH), and physical random access channel (PRACH) are defined as uplink physical channels, and the demodulation reference signal (DMRS) for uplink control / data signals and the sounding reference signal (SRS) used for uplink channel measurement are defined.

[0058] In this specification, PDCCH (Physical Downlink Control Channel) refers to a set of time-frequency resources (e.g., resource elements (REs)) carrying DCI (Downlink Control Information), and PDSCH (Physical Downlink Shared Channel) refers to a set of time-frequency resources carrying downlink data. Additionally, PUCCH (Physical Uplink Control Channel), PUSCH (Physical Uplink Shared Channel), and PRACH (Physical Random Access Channel) respectively refer to sets of time-frequency resources carrying UCI (Uplink Control Information), uplink data, and random access signals. In the following, the expression that a user device transmits / receives PUCCH / PUSCH / PRACH is used in the same sense as transmitting / receiving uplink control information / uplink data / random access signals on or through PUCCH / PUSCH / PRACH, respectively. In addition, the expression that BS transmits / receives PBCH / PDCCH / PDSCH is used with the same meaning as transmitting broadcast information / downlink control information / downlink data on or through PBCH / PDCCH / PDSCH, respectively.

[0059] In this specification, a radio resource (e.g., time-frequency resource) scheduled or set by a BS for a UE for the transmission or reception of PUCCH / PUSCH / PDSCH is also referred to as a PUCCH / PUSCH / PDSCH resource.

[0060] Since the communication device receives a synchronization signal (SS), DMRS, CSI-RS, PRS, PBCH, PDCCH, PDSCH, PUSCH, and / or PUCCH in the form of radio signals on the cell, it is not possible to selectively receive only radio signals containing only a specific physical channel or a specific physical signal through the RF receiver, or to selectively receive only radio signals excluding only a specific physical channel or a specific physical signal through the RF receiver. In actual operation, the communication device first receives radio signals on the cell through the RF receiver, converts the radio signals, which are RF band signals, into baseband signals, and uses one or more processors to decode the physical signals and / or physical channels within the baseband signals. Accordingly, in some implementations of this specification, not receiving a physical signal and / or a physical channel may actually mean that the communication device does not receive wireless signals containing the physical signal and / or physical channel at all, but rather does not attempt to recover the physical signal and / or physical channel from the wireless signals, for example, not attempt to decode the physical signal and / or physical channel.

[0061] As more communication devices require larger communication capacities, the need for improved mobile broadband communication compared to existing radio access technology (RAT) is emerging. Furthermore, massive MTC, which connects multiple devices and objects to provide various services anytime and anywhere, is also one of the major issues to be considered in next-generation communication. In addition, communication system designs that consider reliability and latency-sensitive services / UEs are being discussed. Accordingly, the introduction of next-generation RATs that consider advanced mobile broadband communication, massive MTC, and URLLC (Ultra-Reliable and Low Latency Communication) is being discussed. Currently, 3GPP is conducting studies on next-generation mobile communication systems following the EPC. For convenience, this specification refers to the technology as new RAT (NR) or 5G RAT, and systems that use or support NR are referred to as NR systems.

[0062] FIG. 1 illustrates an example of a communication system 1 to which the implementations of the present specification apply. Referring to FIG. 1, the communication system (1) to which the present specification applies includes a wireless device, a BS, and a network. Here, the wireless device refers to a device that performs communication using wireless access technology (e.g., 5G NR (New RAT), LTE (e.g., E-UTRA)) and may be referred to as a communication / wireless / 5G device. Although not limited thereto, the wireless device may include a robot (100a), a vehicle (100b-1, 100b-2), an XR (eXtended Reality) device (100c), a hand-held device (100d), a home appliance (100e), an IoT (Internet of Thing) device (100f), and an AI device / server (400). For example, the vehicle may include a vehicle equipped with wireless communication capabilities, an autonomous vehicle, a vehicle capable of performing inter-vehicle communication, etc. Here, vehicles may include UAVs (Unmanned Aerial Vehicles) (e.g., drones). XR devices include AR (Augmented Reality) / VR (Virtual Reality) / MR (Mixed Reality) devices and may be implemented in the form of HMDs (Head-Mounted Devices), HUDs (Head-Up Displays) equipped in vehicles, televisions, smartphones, computers, wearable devices, home appliances, digital signage, vehicles, robots, etc. Portable devices may include smartphones, smartpads, wearable devices (e.g., smartwatches, smart glasses), computers (e.g., laptops, etc.). Home appliances may include TVs, refrigerators, washing machines, etc. IoT devices may include sensors, smart meters, etc. For example, BS and networks may be implemented as wireless devices, and specific wireless devices may operate as BS / network nodes to other wireless devices.

[0063] Wireless devices (100a to 100f) can be connected to a network (300) via a BS (200). Artificial Intelligence (AI) technology may be applied to the wireless devices (100a to 100f), and the wireless devices (100a to 100f) can be connected to an AI server (400) via the network (300). The network (300) can be configured using a 3G network, a 4G (e.g., LTE) network, or a 5G (e.g., NR) network. The wireless devices (100a to 100f) may communicate with each other via the BS (200) / network (300), but they may also communicate directly (e.g., sidelink communication) without using the BS / network. For example, vehicles (100b-1, 100b-2) can communicate directly (e.g., V2V (Vehicle to Vehicle) / V2X (Vehicle to everything) communication). Also, IoT devices (e.g., sensors) can communicate directly with other IoT devices (e.g., sensors) or other wireless devices (100a to 100f).

[0064] Wireless communication / connection (150a, 150b) can be established between wireless devices (100a~100f) / BS (200) and BS (200) / wireless devices (100a~100f). Here, the wireless communication / connection can be established through various wireless access technologies (e.g., 5G NR), such as uplink / downlink communication (150a) and sidelink communication (150b) (or D2D communication). Through the wireless communication / connection (150a, 150b), wireless devices and BS / wireless devices can transmit / receive wireless signals to / from each other. To this end, based on various proposals of this specification, at least some of the following may be performed: a process for setting various configuration information for transmitting / receiving wireless signals, a process for various signal processing (e.g., channel encoding / decoding, modulation / demodulation, resource mapping / demapping, etc.), and a resource allocation process.

[0065] FIG. 2 is a block diagram illustrating examples of communication devices capable of performing the method according to the present specification. Referring to FIG. 2, a first wireless device (100) and a second wireless device (200) can transmit and / or receive wireless signals through various wireless access technologies (e.g., LTE, NR). Here, {the first wireless device (100), the second wireless device (200)} may correspond to {wireless device (100x), BS (200)} and / or {wireless device (100x), wireless device (100x)} of FIG. 1.

[0066] The first wireless device (100) includes one or more processors (102) and one or more memories (104), and may additionally include one or more transceivers (106) and / or one or more antennas (108). The processor (102) controls the memory (104) and / or transceivers (106) and may be configured to implement the functions, procedures and / or methods described / suggested below. For example, the processor (102) may process information within the memory (104) to generate a first information / signal and then transmit a wireless signal containing the first information / signal through the transceiver (106). Additionally, the processor (102) may receive a wireless signal containing a second information / signal through the transceiver (106) and then store information obtained from the signal processing of the second information / signal in the memory (104). Memory (104) may be connected to the processor (102) and may store various information related to the operation of the processor (102). For example, memory (104) may store software code containing instructions for performing some or all of the processes controlled by the processor (102) or for performing the procedures and / or methods described / suggested below. Here, the processor (102) and memory (104) may be part of a communication modem / circuit / chip designed to implement wireless communication technology (e.g., LTE, NR). A transceiver (106) may be connected to the processor (102) and may transmit and / or receive wireless signals through one or more antennas (108). The transceiver (106) may include a transmitter and / or receiver. The transceiver (106) may be interchangeably used with an RF (Radio Frequency) unit. In this specification, a wireless device may mean a communication modem / circuit / chip.

[0067] According to one example, the first wireless device or terminal (100) may include a processor (102) connected to the RF transceiver and a memory (104). The memory (104) may include at least one program capable of performing operations related to the embodiments described in FIGS. 19 to 21 below.

[0068] Specifically, the processor (102) controls the RF transceiver (106) to transmit a signal for data transmission and sensing to a base station, receives a reflected signal of the signal to acquire first radar data for a first frequency band, and can receive second radar data from the base station including synthetic radar data generated based on the first radar data. Here, the synthetic radar data may be radar data synthesized based on the first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

[0069] Alternatively, a processing device may be configured including a processor (102) and a memory (104). In this case, the processing device may include at least one processor; and at least one memory connected to the at least one processor and storing instructions, wherein the instructions, based on execution by the at least one processor, may cause the UE (100) to: transmit a signal for data transmission and sensing to a base station, receive a reflected signal of the signal to acquire first radar data for a first frequency band, and receive second radar data including synthetic radar data generated based on the first radar data from the base station. Here, the synthetic radar data may be radar data synthesized based on the first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

[0070] The second wireless device (200) includes one or more processors (202) and one or more memories (204), and may additionally include one or more transceivers (206) and / or one or more antennas (208). The processor (202) controls the memory (204) and / or transceivers (206) and may be configured to implement the functions, procedures and / or methods described / suggested below. For example, the processor (202) may process information within the memory (204) to generate a third information / signal and then transmit a wireless signal containing the third information / signal through the transceiver (206). Additionally, the processor (202) may receive a wireless signal containing a fourth information / signal through the transceiver (206) and then store information obtained from the signal processing of the fourth information / signal in the memory (204). Memory (204) may be connected to the processor (202) and may store various information related to the operation of the processor (202). For example, memory (204) may store software code containing instructions for performing some or all of the processes controlled by the processor (202) or for performing the procedures and / or methods described / suggested below. Here, the processor (202) and memory (204) may be part of a communication modem / circuit / chip designed to implement wireless communication technology (e.g., LTE, NR). A transceiver (206) may be connected to the processor (202) and may transmit and / or receive wireless signals through one or more antennas (208). The transceiver (206) may include a transmitter and / or receiver. The transceiver (206) may be interchangeable with an RF unit. In this specification, a wireless device may mean a communication modem / circuit / chip.

[0071] The wireless communication technology implemented in the wireless device (100, 200) of this specification may include LTE, NR, and 6G, as well as Narrowband Internet of Things for low-power communication. In this case, for example, NB-IoT technology may be an example of LPWAN (Low Power Wide Area Network) technology and may be implemented according to standards such as LTE Cat NB1 and / or LTE Cat NB2, but is not limited to the names mentioned above. Additionally, or generally, the wireless communication technology implemented in the wireless device (XXX, YYY) of this specification may perform communication based on LTE-M technology. In this case, for example, LTE-M technology may be an example of LPWAN technology and may be referred to by various names such as eMTC (enhanced Machine Type Communication). For example, LTE-M technology may be implemented in at least one of various standards such as 1) LTE CAT 0, 2) LTE Cat M1, 3) LTE Cat M2, 4) LTE non-BL (non-Bandwidth Limited), 5) LTE-MTC, 6) LTE Machine Type Communication, and / or 7) LTE M, and is not limited to the names mentioned above. Additionally or generally, wireless communication technology implemented in the wireless device (XXX, YYY) of this specification may include at least one of ZigBee, Bluetooth, and Low Power Wide Area Network (LPWAN) with consideration for low-power communication, and is not limited to the names mentioned above. As an example, ZigBee technology can create personal area networks (PANs) related to small / low-power digital communication based on various standards such as IEEE 802.15.4, and may be referred to by various names.

[0072] Hereinafter, hardware elements of the wireless device (100, 200) will be described in more detail. Although not limited thereto, one or more protocol layers may be implemented by one or more processors (102, 202). For example, one or more processors (102, 202) may implement one or more layers (e.g., functional layers such as a physical (PHY) layer, a medium access control (MAC) layer, a radio link control (RLC) layer, a packet data convergence protocol (PDCP) layer, a radio resource control (RRC) layer, and a service data adaptation protocol (SDAP). One or more processors (102, 202) may generate one or more protocol data units (PDU) and / or one or more service data units (SDU) according to the functions, procedures, proposals and / or methods disclosed herein. One or more processors (102, 202) may generate messages, control information, data, or information according to the functions, procedures, proposals, and / or methods disclosed in this specification. One or more processors (102, 202) may generate a signal (e.g., baseband signal) containing a PDU, SDU, message, control information, data, or information according to the functions, procedures, proposals, and / or methods disclosed in this specification and provide it to one or more transceivers (106, 206). One or more processors (102, 202) may receive a signal (e.g., baseband signal) from one or more transceivers (106, 206) and may obtain a PDU, SDU, message, control information, data, or information according to the functions, procedures, proposals, and / or methods disclosed in this specification.

[0073] One or more processors (102, 202) may be referred to as a controller, microcontroller, microprocessor, or microcomputer. One or more processors (102, 202) may be implemented by hardware, firmware, software, or a combination thereof. For example, one or more Application Specific Integrated Circuits (ASICs), one or more Digital Signal Processors (DSPs), one or more Digital Signal Processing Devices (DSPDs), one or more Programmable Logic Devices (PLDs), or one or more Field Programmable Gate Arrays (FPGAs) may be included in one or more processors (102, 202). The functions, procedures, proposals, and / or methods disclosed herein may be implemented using firmware or software, and the firmware or software may be implemented to include modules, procedures, functions, etc. Firmware or software configured to perform the functions, procedures, proposals, and / or methods disclosed in this specification may be included in one or more processors (102, 202) or stored in one or more memories (104, 204) and driven by one or more processors (102, 202). The functions, procedures, proposals, and / or methods disclosed in this specification may be implemented using firmware or software in the form of code, instructions, and / or sets of instructions.

[0074] One or more memories (104, 204) may be connected to one or more processors (102, 202) and may store various forms of data, signals, messages, information, programs, codes, instructions, and / or commands. One or more memories (104, 204) may be composed of ROM, RAM, EPROM, flash memory, hard drive, registers, cache memory, computer read storage media, and / or combinations thereof. One or more memories (104, 204) may be located inside and / or outside of one or more processors (102, 202). Additionally, one or more memories (104, 204) may be connected to one or more processors (102, 202) through various technologies such as wired or wireless connections.

[0075] One or more transceivers (106, 206) may transmit user data, control information, wireless signals / channels, etc., as mentioned in the methods and / or operation flowcharts, etc., of this specification to one or more other devices. One or more transceivers (106, 206) may receive user data, control information, wireless signals / channels, etc., as mentioned in the functions, procedures, proposals, methods and / or operation flowcharts, etc., disclosed in this specification from one or more other devices. For example, one or more transceivers (106, 206) may be connected to one or more processors (102, 202) and may transmit and / or receive wireless signals. For example, one or more processors (102, 202) may control one or more transceivers (106, 206) to transmit user data, control information, or wireless signals to one or more other devices. Additionally, one or more processors (102, 202) may control one or more transceivers (106, 206) to receive user data, control information, or wireless signals from one or more other devices. Additionally, one or more transceivers (106, 206) may be connected to one or more antennas (108, 208), and one or more transceivers (106, 206) may be configured to transmit and / or receive user data, control information, wireless signals / channels, etc., as mentioned in the functions, procedures, proposals, methods, and / or operation flowcharts disclosed herein through one or more antennas (108, 208). In this specification, one or more antennas may be a plurality of physical antennas or a plurality of logical antennas (e.g., antenna ports). One or more transceivers (106, 206) can convert the received wireless signal / channel, etc. from an RF band signal to a baseband signal in order to process the received user data, control information, wireless signal / channel, etc. using one or more processors (102, 202).One or more transceivers (106, 206) can convert user data, control information, wireless signals / channels, etc. processed using one or more processors (102, 202) from baseband signals to RF band signals. To this end, one or more transceivers (106, 206) may include (analog) oscillators and / or filters.

[0076] FIG. 3 illustrates another example of a wireless device capable of performing the implementation(s) of the present specification. Referring to FIG. 3, the wireless device (100, 200) corresponds to the wireless device (100, 200) of FIG. 2 and may be composed of various elements, components, units / parts, and / or modules. For example, the wireless device (100, 200) may include a communication unit (110), a control unit (120), a memory unit (130), and additional elements (140). The communication unit may include a communication circuit (112) and transceiver(s) (114). For example, the communication circuit (112) may include one or more processors (102, 202) and / or one or more memories (104, 204) of FIG. 2. For example, the transceiver(s) (114) may include one or more transceivers (106, 206) and / or one or more antennas (108, 208) of FIG. 2. The control unit (120) is electrically connected to the communication unit (110), the memory unit (130), and additional elements (140) and controls the general operation of the wireless device. For example, the control unit (120) may control the electrical / mechanical operation of the wireless device based on a program / code / command / information stored in the memory unit (130). Additionally, the control unit (120) may transmit information stored in the memory unit (130) to the outside (e.g., another communication device) via a wireless / wired interface through the communication unit (110), or store information received from the outside (e.g., another communication device) via a wireless / wired interface through the communication unit (110) in the memory unit (130).

[0077] The additional element (140) may be configured in various ways depending on the type of wireless device. For example, the additional element (140) may include at least one of a power unit / battery, an input / output unit (I / O unit), a driving unit, and a computing unit. Although not limited thereto, the wireless device may be implemented in the form of a robot (Fig. 1, 100a), a vehicle (Fig. 1, 100b-1, 100b-2), an XR device (Fig. 1, 100c), a portable device (Fig. 1, 100d), a home appliance (Fig. 1, 100e), an IoT device (Fig. 1, 100f), a UE for digital broadcasting, a holographic device, a public safety device, an MTC device, a medical device, a fintech device (or financial device), a security device, a climate / environment device, an AI server / device (Fig. 1, 400), a BS (Fig. 1, 200), a network node, etc. Depending on the use—e.g., service—the wireless device may be movable or used in a fixed location.

[0078] In FIG. 3, various elements, components, units / parts, and / or modules within the wireless device (100, 200) may be entirely interconnected via a wired interface, or at least a portion may be wirelessly connected via a communication unit (110). For example, within the wireless device (100, 200), the control unit (120) and the communication unit (110) may be wired, and the control unit (120) and the first unit (e.g., 130, 140) may be wirelessly connected via the communication unit (110). Additionally, each element, component, unit / part, and / or module within the wireless device (100, 200) may include one or more additional elements. For example, the control unit (120) may be composed of one or more sets of processors. For example, the control unit (120) may be composed of a set of a communication control processor, an application processor, an Electronic Control Unit (ECU), a graphics processing processor, a memory control processor, etc. As another example, the memory unit (130) may be composed of RAM (Random Access Memory), DRAM (Dynamic RAM), ROM (Read Only Memory), flash memory, volatile memory, non-volatile memory, transitory memory, non-transitory memory, and / or a combination thereof.

[0079] In this specification, at least one memory (e.g., 104 or 204) may store instructions or programs, and said instructions or programs may, when executed, cause at least one processor operablely connected to said at least one memory to perform operations according to some embodiments or implementations of this specification.

[0080] In this specification, a computer-readable (non-volatile or non-transient) storage medium may store at least one instruction or computer program, and when executed by at least one processor, said at least one instruction or computer program may cause said at least one processor to perform operations according to some embodiments or implementations of this specification.

[0081] In this specification, a processing device or apparatus may include at least one processor and at least one computer memory connectable to said at least one processor. said at least one computer memory may store instructions or programs, and said instructions or programs, when executed, may cause at least one processor operablely connected to said at least one memory to perform operations according to some embodiments or implementations of this specification.

[0082] In this specification, a computer program may include program code that is stored on at least one computer-readable (non-volatile) storage medium and, when executed, performs operations according to some implementations of this specification or causes at least one processor to perform operations according to some implementations of this specification. The computer program may be provided in the form of a computer program product. The computer program product may include at least one computer-readable (non-volatile) storage medium.

[0083] A communication device of the present specification comprises at least one processor; and at least one computer memory operably connected to the at least one processor and storing instructions that, when executed, cause the at least one processor to perform operations according to the examples(s) of the present specification described below.

[0084] Figure 4 illustrates an example of a frame structure available in a 3GPP-based wireless communication system.

[0085] The structure of the frame in Fig. 4 is merely an example, and the number of subframes, slots, and symbols in the frame can be varied. In an NR system, OFDM numerology (e.g., subcarrier spacing (SCS)) may be set differently among multiple cells aggregated to a single UE. Accordingly, the (absolute time) duration of a time resource (e.g., subframe, slot, or transmission time interval (TTI)) consisting of the same number of symbols may be set differently among the aggregated cells. Here, symbols may include OFDM symbols (or cyclic prefix-orthogonal frequency division multiplexing (CP-OFDM) symbols) and SC-FDMA symbols (or discrete Fourier transform-spread-OFDM (DFT-s-OFDM) symbols). In this specification, symbols, OFDM-based symbols, OFDM symbols, CP-OFDM symbols, and DFT-s-OFDM symbols refer to They can be substituted for each other.

[0086] Referring to Fig. 4, uplink and downlink transmissions in an NR system are organized into frames. Each frame is T f= (△f max *N f / 100)*T c = It has a duration of 10 ms and is divided into two half-frames, each with a duration of 5 ms. Here, T is the basic time unit for NR. c = 1 / (△f max *N f ) and, △f max = 480*10 3 It is Hz, and N f = 4096. For reference, T is the standard time unit for LTE. s = 1 / (△f ref *N f,ref ) and, △f ref = 15*10 3 It is Hz, and N f,ref =2048. T s Wow T c is a constant κ = T s / T c It has a relationship of = 64. Each half-frame consists of 5 subframes, and the period T of a single subframe. sf is 1ms. Subframes are further divided into slots, and the number of slots within a subframe depends on the subcarrier spacing. Each slot consists of 14 or 12 OFDM symbols based on a cyclic prefix. For a normal cyclic prefix (CP), each slot consists of 14 OFDM symbols, while for an extended CP, each slot consists of 12 OFDM symbols. The above numerology is an exponentially scalable subcarrier spacing △f = 2 u It depends on 15 kHz. The following table shows the subcarrier spacing △f = 2 for normalized CP. u *Number of OFDM symbols per slot according to 15 kHz (N slot symb ), number of slots per frame (N frame,uslot ) and the number of slots per subframe (N subframe,u slot It represents ).

[0087]

[0088] The following table shows the subcarrier spacing △f = 2 for extended CP. u This shows the number of OFDM symbols per slot, the number of slots per frame, and the number of slots per subframe according to *15 kHz.

[0089]

[0090] For a subcarrier interval setting u, the slots are arranged in increasing order n within the subframe. u s ∈ {0, ..., nsubframe,u slot - 1} and n in increasing order within the frame u s,f ∈ {0, ..., n frame,u slot - 1} is numbered.

[0091] A slot contains multiple (e.g., 14 or 12) symbols in the time domain. For each numeral (e.g., subcarrier interval) and carrier, a common resource block (CRB) N indicated by upper-layer signaling (e.g., radio resource control (RRC) signaling) start,u grid Starting from,N size,u grid,x *N RB sc individual subcarriers and N subframe,u symb A resource grid of N OFDM symbols is defined. Here, N size,u grid,x is the number of resource blocks (RB) in the resource grid, and the subscript x is DL for downlinks and UL for uplinks. RBsc is the number of subcarriers per RB, and in 3GPP-based wireless communication systems, N RB sc is typically 12. There is one resource grid for a given antenna port p, subcarrier spacing configuration u, and transmission direction (DL or UL). Carrier bandwidth N for subcarrier spacing configuration u. size,u grid This is given to the UE by upper-layer parameters (e.g., RRC parameters) from the network. Each element within the resource grid for antenna port p and subcarrier spacing u is referred to as a resource element (RE), and one complex symbol can be mapped to each resource element. Each resource element within the resource grid is uniquely identified by an index k in the frequency domain and an index l indicating the symbol position relative to a reference point in the time domain. In an NR system, RBs are defined by 12 consecutive subcarriers in the frequency domain. In an NR system, RBs can be classified into Common Resource Blocks (CRBs) and Physical Resource Blocks (PRBs). CRBs are numbered upwards from 0 in the frequency domain for subcarrier spacing u. The center of subcarrier 0 of CRB 0 for subcarrier spacing u coincides with 'Point A', which is the common reference point for the resource block grids. The PRBs for the subcarrier spacing setting u are defined within the bandwidth part (BWP), and range from 0 to N size,u BWP,i Numbered up to -1, where i is the number of the above bandwidth part. Common resource block n u CRB and physical resource block n within bandwidth part i PRB The relationships between them are as follows: n u PRB = n uCRB +N start,u BWP,i , here N start,u BWP,i is a common resource block where the above bandwidth part starts relative to CRB 0. A BWP contains multiple consecutive RBs in the frequency domain. For example, a BWP is a given numerator u within a BWP i on a given carrier. i It is a subset of contiguous CRBs defined for. The carrier may contain up to N (e.g., 5) BWPs. A UE may be configured to have one or more BWPs on a given component carrier. Data communication is performed through the enabled BWPs, and only a predetermined number (e.g., 1) of the BWPs configured for the UE may be enabled on the carrier.

[0092] For each serving cell within a set of DL BWPs or UL BWPs, the network establishes at least an initial DL BWP and one initial UL BWP (if the serving cell is configured with an uplink) or two initial UL BWPs (if using a supplementary uplink). The network may also establish additional ULs and DL BWPs for the serving cell. For each DL BWP or UL BWP, the UE is provided with the following parameters for the serving cell: i) subcarrier spacing, ii) circular prefix, iii) N start BWP Assuming = 275, offset RB set and length L RB CRBN provided by the RRC parameter locationAndBandwidth, which indicates as the resource indicator value (RIV). start BWP =O carrier +RB start and the number of contiguous RBs N size BWP =LRB , and O provided by the RRC parameter offsetToCarrier for the subcarrier spacing carrier ; Index within the set of the above DL BWPs or UL BWPs; set of BWP-common parameters and set of BWP-exclusive parameters.

[0093] Virtual resource blocks (VRBs) are defined within the bandwidth part and range from 0 to N size,u BWP,i Numbered up to -1, where i is the number of the above bandwidth part. VRBs are mapped to physical resource blocks (PRBs) according to interleaved or non-interleaved mapping. In some implementations, for non-interleaved VRB-to-PRB mapping, VRB n can be mapped to PRB n.

[0094] NR frequency bands are defined as two types of frequency ranges, FR1 and FR2, where FR2 is also referred to as millimeter wave (mmW). The following table illustrates the frequency ranges in which NR can operate.

[0095] Frequency Range designationCorresponding frequency rangeSubcarrier SpacingFR1410MHz - 7125MHz15, 30, 60kHzFR224250MHz - 52600MHz60, 120, 240kHz

[0096] Figure 5 illustrates physical channels used in a 3GPP-based communication system, which is an example of a wireless communication system, and the signal transmission / reception process using them.

[0097] A UE that has been turned on again after being turned off or has lost connection with a wireless communication system first performs an initial cell search process, such as searching for a suitable cell to camp on and synchronizing with said cell or the BS of said cell (S11). During the initial cell search process, the UE receives a synchronization signal block (SSB) (also called an SSB / PBCH block) from the BS. The SSB includes a primary synchronization signal (PSS), a secondary synchronization signal (SSS), and a physical broadcast channel (PBCH). Based on the PSS / SSS, the UE synchronizes with the BS and obtains information such as the cell identifier (ID). Additionally, the UE can obtain broadcast information within the cell based on the PBCH. Meanwhile, during the initial cell search process, the UE can receive a downlink reference signal (DL RS) to check the downlink channel status.

[0098] After completing the initial cell search, the UE can camp on the cell. After camping on the cell, the UE monitors the PDCCH on the cell and receives the PDSCH according to the downlink control information (DCI) carried by the PDCCH to obtain more specific system information (S12).

[0099] Subsequently, the UE may perform a random access procedure to complete access to the BS (S13 to S16). For example, during the random access procedure, the UE may transmit a preamble through a physical random access channel (PRACH) (S13) and receive a random access response (RAR) for the preamble through a PDCCH and a corresponding PDSCH (S14). If the reception of the RAR for the UE fails, the UE may attempt to re-transmit the preamble. In the case of contention-based random access, a contention resolution procedure (S16) may be performed, which includes transmitting a PUSCH based on the UL resource allocation included in the RAR (S15) and receiving a PDCCH and a corresponding PDSCH.

[0100] A UE that has performed the procedure described above may subsequently perform the reception of PDCCH / PDSCH (S17) and the transmission of PUSCH / PUCCH (S18) as part of a general uplink / downlink signal transmission process. The control information transmitted by the UE to the BS is collectively referred to as uplink control information (UCI). UCI includes HARQ ACK / NACK (Hybrid Automatic Repeat and reQuest Acknowledgement / Negative-ACK) (also called HARQ-ACK), scheduling request (SR), channel state information (CSI), etc. CSI may include channel quality indicator (CQI), precoding matrix indicator (PMI), and / or rank indicator, etc. UCI is generally transmitted via PUCCH, but may be transmitted via PUSCH if control information and traffic data need to be transmitted simultaneously. In addition, the UE can transmit UCI atypically via PUSCH based on network requests / instructions.

[0101] FIG. 6 illustrates an arbitrary connection process that may be applied to the implementation(s) of the present specification. In particular, FIG. 6(a) illustrates a four-step arbitrary connection process, and FIG. 6(b) illustrates a two-step arbitrary connection process.

[0102] The random access process can be used for various purposes, such as initial access, uplink adjustment, resource allocation, handover, reconfiguration after a wireless link failure, and location measurement. Random access processes are classified into contention-based and dedicated (i.e., non-contention-based) processes. Contention-based random access processes are generally used for initial access, while dedicated random access processes are used for handover, when downlink data reaches the network, and when reconfiguring uplink adjustment for location measurement. In a contention-based random access process, the UE randomly selects a random access (RA) preamble. Therefore, it is possible for multiple UEs to transmit the same RA preamble simultaneously, which necessitates a subsequent contention resolution process. In contrast, in a dedicated random access process, the UE uses an RA preamble uniquely assigned to it by the BS. Consequently, the UE can perform the random access process without conflicts with other UEs.

[0103] Referring to FIG. 6(a), the contention-based random access process includes the following four steps. Hereinafter, the messages transmitted in steps 1 through 4 may be referred to as Msg1 through Msg4, respectively.

[0104] - Step 1: The UE transmits the RA preamble via PRACH.

[0105] - Step 2: The UE receives a random access response (RAR) from the BS via PDSCH.

[0106] - Step 3: The UE transmits UL data to the BS via PUSCH based on RAR. Here, the UL data includes Layer 2 and / or Layer 3 messages.

[0107] - Step 4: The UE receives a contention resolution message from the BS via PDSCH.

[0108] The UE can receive information regarding random access from the BS through system information. For example, information regarding RACH times associated with SSBs on the cell may be provided through system information. The UE can select an SSB among those received on the cell for which the reference signal received power (RSRP) measured based on the SSB exceeds a threshold, and transmit an RA preamble through the PRACH associated with the selected SSB. For example, if random access is required, the UE transmits Msg1 (e.g., preamble) to the BS on the PRACH. The BS can distinguish each random access preamble through the time / frequency resource (RA Occasion, RO) and the random access preamble index (Preamble Index, PI). When the BS receives a random access preamble from the UE, the BS transmits a RAR message to the UE on the PDSCH. To receive a RAR message, the UE monitors a CRC-masked L1 / L2 control channel (PDCCH) containing scheduling information for a RAR message within a preset time window, a RAR window (e.g., ra-ResponseWindow), which is a Random Access-RNTI (RA-RNTI). The length of the RAR window may be set by upper-level signaling, and the RAR window may start at a specific timing after a PRACH transmission (e.g., the first symbol of the fastest control resource set (CORESET) of the Type-1 PDCCH common seek space, starting at least one symbol after the PRACH time corresponding to the PRACH transmission). When scheduling information is received through the PDCCH masked by the RA-RNTI, the UE may receive a RAR message from the PDSCH indicated by the scheduling information.Subsequently, the UE determines whether there is a RAR for itself within the aforementioned RAR message. Whether a RAR for itself exists can be verified by checking whether a RAPID (Random Access preamble ID) exists for the preamble transmitted by the UE. The index of the preamble transmitted by the UE and the RAPID may be the same. A RAR includes a corresponding random access preamble index, timing offset information for UL synchronization (e.g., a timing advance command (TAC)), UL scheduling information for sending Msg3 (e.g., a UL grant), and temporary UE identification information (e.g., Temporary-C-RNTI, TC-RNTI). Upon receiving the RAR, the UE sends Msg3 via PUSCH according to the UL scheduling information and timing offset values ​​within the RAR. Msg3 may include the UE's ID (or the UE's global ID). Additionally, Msg3 may include information related to an RRC connection request for initial access to the network (e.g., an RRCSetupRequest message). After receiving Msg3, the BS sends Msg4, a contention resolution message, to the UE. If the UE receives the contention resolution message and the contention is successfully resolved, TC-RNTI is changed to C-RNTI. Msg4 includes the UE's ID and / or It may include information related to the RRC connection (e.g., RRCSetup message). If the information transmitted via Msg3 does not match the information received via Msg4, or if Msg4 is not received for a certain period of time, the UE may report that the contention resolution failed and retransmit Msg3.

[0109] Meanwhile, the dedicated random access process includes the following three steps. Hereinafter, the messages transmitted in steps 0 to 2 may be referred to as Msg0 to Msg2, respectively. The dedicated random access process may be triggered by the UE by the BS using a PDCCH (hereinafter referred to as the PDCCH order) for commanding the transmission of an RA preamble.

[0110] - Step 0: BS assigns the RA preamble to the UE via dedicated signaling.

[0111] - Step 1: The UE transmits the RA preamble via PRACH.

[0112] - Step 2: The UE receives the RAR via the PDSCH from the BS.

[0113] The operations of steps 1 to 2 of a dedicated random access process may be the same as steps 1 to 2 of a contention-based random access process.

[0114] In NR systems, lower latency than in existing systems may be required. Additionally, a four-stage random access process may not be desirable, particularly for latency-sensitive services such as URLLC. A low-latency random access process may be required within various scenarios of NR systems. When the implementation(s) of this specification are performed with a random access process, to reduce latency in the random access process, the implementation(s) of this specification may be performed with the following two-stage random access process.

[0115] Referring to FIG. 6(b), the two-stage random access process may consist of two stages: the transmission of MsgA from the UE to the BS and the transmission of MsgB from the BS to the UE. The transmission of MsgA may include the transmission of an RA preamble via PRACH and the transmission of a UL payload via PUSCH. In the transmission of MsgA, PRACH and PUSCH may be transmitted using time division multiplexing (TDM). Alternatively, in the transmission of MsgA, PRACH and PUSCH may be transmitted using frequency division multiplexing (FDM).

[0116] A BS that receives MsgA may transmit MsgB to a UE. MsgB may include a RAR for said UE. After MsgA transmission, said UE monitors for a response from the network within a time window to monitor for a RAR for a two-stage random access process. The length of said time window may be set by upper-layer signaling, and said time window may start at a specific timing after MsgA transmission (e.g., the first symbol of the fastest CORESET of the Type-1 PDCCH common seek space starting at least one symbol after the last symbol of the PUCCH time corresponding to the PRACH transmission of said MsgA transmission).

[0117] A message related to an RRC connection request (e.g., RRCSetupRequest message) requesting to establish a connection between the RRC layer of the BS and the RRC layer of the UE may be transmitted by being included in the payload of MsgA. In this case, MsgB may be used to transmit RRC connection-related information (e.g., RRCSetup message). Alternatively, the RRC connection request message (e.g., RRCSetupRequest message) may be transmitted via PUSCH transmitted based on a UL grant within MsgB. In this case, the RRC connection-related information (e.g., RRCSetup message) related to the RRC connection request may be transmitted via PDSCH associated with said PUSCH transmission after the PUSCH transmission based on MsgB.

[0118] Below, physical channels that can be used in 3GPP-based wireless communication systems are described in more detail.

[0119] A PDCCH carries a DCI. For example, a PDCCH (i.e., a DCI) carries the transmission format and resource allocation of the downlink shared channel (DL-SCH), resource allocation information for the uplink shared channel (UL-SCH), paging information for the paging channel (PCH), system information on the DL-SCH, resource allocation information for control messages of the layer above the physical layer (hereinafter referred to as the upper layer) among the protocol stacks of the UE / BS, such as random access response (RAR) transmitted on the PDSCH, transmission power control commands, and the activation / deactivation of configured scheduling (CS). A DCI containing resource allocation information for the DL-SCH is also called a PDSCH scheduling DCI, and a DCI containing resource allocation information for the UL-SCH is also called a PUSCH scheduling DCI. The DCI includes a cyclic redundancy check (CRC), and the CRC is masked / scrambled with various identifiers (e.g., radio network temporary identifier (RNTI)) depending on the owner or use of the PDCCH. For example, if the PDCCH is for a specific UE, the CRC is masked with the UE identifier (e.g., cell RNTI (C-RNTI)). If the PDCCH is for paging, the CRC is masked with the paging RNTI (P-RNTI). If the PDCCH is for system information (e.g., system information block (SIB)), the CRC is masked with the system information RNTI (SI-RNTI). If the PDCCH is for a random access response, the CRC is masked with the random access RNTI (RA-RATI).

[0120] The scheduling of a PDCCH on one serving cell to a PDSCH or PUSCH on another serving cell is called cross-carrier scheduling. Cross-carrier scheduling using a carrier indicator field (CIF) may allow a PDCCH on a serving cell to schedule resources on another serving cell. Meanwhile, the scheduling of a PDSCH or PUSCH on a serving cell to a serving cell is called self-carrier scheduling. When cross-carrier scheduling is used in a cell, the BS may provide the UE with information regarding the cell scheduling said cell. For example, the BS may provide the UE with whether the serving cell is scheduled by a PDCCH on another (scheduling) cell or by said serving cell, and if said serving cell is scheduled by another (scheduling) cell, which cell signals downlink assignments and uplink grants for said serving cell. In this specification, a cell carrying a PDCCH is referred to as a scheduling cell, and a cell in which the transmission of a PUSCH or PDSCH is scheduled by a DCI included in the PDCCH, that is, a cell carrying a PUSCH or PDSCH scheduled by the PDCCH, is referred to as a scheduled cell.

[0121] PDSCH is a physical layer DL channel for DL ​​data transport. PDSCH carries downlink data (e.g., DL-SCH transport blocks) and applies modulation methods such as QPSK (Quadrature Phase Shift Keying), 16 QAM (Quadrature Amplitude Modulation), 64 QAM, and 256 QAM. Codewords are generated by encoding transport blocks (TB). PDSCH can carry up to two codewords. Scrambling and modulation mapping are performed for each codeword, and the modulation symbols generated from each codeword can be mapped to one or more layers. Each layer is mapped to a radio resource along with DMRS to generate an OFDM symbol signal, which is then transmitted through the corresponding antenna port.

[0122] The UE must have uplink resources available to it for UL-SCH data transmission and downlink resources available to it for DL-SCH data reception. Uplink resources and downlink resources are assigned to the UE through resource allocation by the BS. Resource allocation may include time domain resource allocation (TDRA) and frequency domain resource allocation (FDRA). In this specification, uplink resource allocation is also referred to as uplink grant, and downlink resource allocation is also referred to as downlink assignment. Uplink grant is dynamically received by the UE on the PDCCH or within the RAR, or is semi-persistently set to the UE by RRC signaling from the BS. Downlink assignment is dynamically received by the UE on the PDCCH, or is semi-persistently set to the UE by RRC signaling from the BS.

[0123] In UL, the BS can dynamically allocate uplink resources to the UE via PDCCH(s) addressed to a cell radio network temporary Identifier (C-RNTI). The UE monitors the PDCCH(s) to identify potential uplink grant(s) for UL transmission. Additionally, the BS can allocate uplink resources to the UE using a configured grant (CG). Two types of configured grants, Type 1 and Type 2, may be used. In the case of Type 1, the BS directly provides the configured uplink grant (including periodicity) via RRC signaling. In the case of Type 2, the BS sets the period of an RRC-configured uplink grant through RRC signaling and can signal and activate or deactivate the said uplink grant through a PDCCH (PDCCH addressed to CS-RNTI) addressed to a configured scheduling RNTI (CS-RNTI). For example, in the case of Type 2, the PDCCH addressed to CS-RNTI indicates that the said uplink grant may be implicitly reused according to the period set by RRC signaling until it is deactivated.

[0124] In DL, the BS can dynamically allocate downlink resources to the UE via PDCCH(s) addressed by C-RNTI. The UE monitors the PDCCH(s) to identify potential downlink assignments. Additionally, the BS can allocate downlink resources to the UE using semi-static scheduling (SPS). The BS can set the period of the configured downlink assignments via RRC signaling and signal and enable or disable the configured downlink assignments via PDCCHs addressed by CS-RNTI. For example, a PDCCH addressed by CS-RNTI indicates that the corresponding downlink assignment may be implicitly reused according to the period set by RRC signaling until it is disabled.

[0125] A control resource set (CORESET), which is a set of time-frequency resources that allows the UE to monitor PDCCH, may be defined and / or configured. The CORESET consists of a set of physical resource blocks (PRBs) having a duration of one to three OFDM symbols. The PRBs constituting the CORESET and the CORESET duration may be provided to the UE through upper layer (e.g., RRC) signaling. Within the configured CORESET(s), a set of PDCCH candidates is monitored according to the corresponding search space sets. As specified herein, monitoring implies decoding (also known as blind decoding) each PDCCH candidate according to the monitored DCI formats.

[0126] The set of PDCCH candidates monitored by the UE is defined in terms of PDCCH search space sets. A search space set can be a common search space (CSS) set or a UE-specific search space (USS) set. Each CORESET setting is associated with one or more search space sets, and each search space set is associated with one CORESET setting.

[0127] A set of PDCCH candidates can be monitored in one or more CORESETs on an active DL BWP on each active serving cell where PDCCH monitoring is configured, wherein monitoring implies receiving each PDCCH candidate and decoding it according to the monitored DCI formats.

[0128] Based on the CORESET / scan space set configuration, the UE can monitor PDCCH candidates from one or more SS sets within the slot. The occasion (e.g., time / frequency resources) when PDCCH candidates must be monitored is defined as a PDCCH (monitoring) time. One or more PDCCH (monitoring) times can be configured within the slot.

[0129] Wireless communication systems are being widely deployed to provide various types of communication services, such as voice and data. There is an increasing demand for higher data rates to accommodate incoming new services and / or scenarios where the virtual and real worlds are intermingled. To handle these endless demands, new communication technologies beyond 5G are required. New communication technologies beyond 6G (hereinafter 6G) systems are intended to provide (i) very high data rates per device, (ii) a very large number of connected devices, (iii) global connectivity, (iv) very low latency, (v) reduced energy consumption of battery-free IoT devices, (vi) ultra-reliable connectivity, and (vii) connected intelligence with machine learning capabilities. The use of the following technologies is being considered in 6G systems: artificial intelligence (AI), terahertz (THX) communication, optical wireless communication (OWC), free space optics (FSO) backhaul networks, massive multiple-input multiple-output (MIMO) technology, blockchain, 3D networking, quantum communication, unmanned aerial vehicles (UAVs), cell-free communication, integration of wireless information and energy transmission, integration of sensing and communication, integration of access backhaul networks, holographic beamforming, big data analytics, and large intelligent surface (LIS).

[0130] In particular, attempts to integrate artificial intelligence (AI) into communication systems are surging. The approaches being attempted in relation to AI can be broadly categorized into AI for communications (AI4C), which utilizes AI to improve communication performance, and communications for AI (C4AI), which advances communication technology to support AI. In the field of AI4C, there are attempts to design systems by replacing the roles of channel encoders / decoders, modulators / demodulators, or channel equalizers with end-to-end autoencoders or neural networks. In the field of C4AI, there is a method utilizing federated learning—a technique of distributed learning—to update a common prediction model while protecting personal information by sharing only the model's weights or gradients with the server, without sharing raw device data. Additionally, there is a method to distribute the load across devices, network edges, and cloud servers through split inference.

[0131] The introduction of AI into communications can streamline and enhance real-time data transmission. AI can use numerous analyses to determine how complex target tasks are performed. In other words, AI can increase efficiency and reduce processing latency.

[0132] Time-consuming tasks such as handover, network selection, and resource scheduling can be performed instantly by using AI. AI can also play a significant role in machine-to-machine, machine-to-human, and human-to-machine communication. AI-based communication systems can be supported by metamaterials, intelligent structures, intelligent networks, intelligent devices, intelligence cognitive radio, self-sustaining wireless networks, and machine learning.

[0133] Until recently, attempts to integrate AI into wireless communication systems have focused on the application layer and network layer, particularly on wireless resource management and allocation. However, research on integrating AI into wireless communication systems is increasingly advancing toward the MAC layer and physical layer, with attempts emerging specifically to combine deep learning with wireless transmission at the physical layer. AI-based physical layer transmission refers to the application of signal processing and communication mechanisms based on AI drivers, rather than traditional communication frameworks, regarding fundamental signal processing and communication mechanisms. Examples include deep learning-based channel coding and decoding, deep learning-based signal estimation and detection, deep learning-based MIMO mechanisms, and AI-based resource scheduling and allocation.

[0134] Machine learning (ML) can be used for channel estimation and channel tracking, and in the physical layer of DL for power allocation, interference cancellation, etc. In addition, machine learning can be used for antenna selection, power control, symbol detection, etc. in MIMO systems.

[0135] However, applying deep neural networks (DNNs) for transmission at the physical layer may have the following problems.

[0136] Deep learning-based AI algorithms require a vast amount of training data to optimize training parameters. However, due to limitations in acquiring training data from specific channel environments, a large amount of offline training data is used. Consequently, static training on training data from specific channel environments can lead to contradictions between the dynamic characteristics and diversity of wireless channels.

[0137] Furthermore, current deep learning primarily targets real signals. However, signals at the physical layer of wireless communication are complex signals. Further research is needed on neural networks that detect complex domain signals to match the characteristics of wireless communication signals.

[0138] Below, we will examine machine learning in more detail.

[0139] Machine learning refers to a series of operations for training machines to create machines capable of performing tasks that humans can or find difficult to do. Machine learning requires data and learning models. Data learning methods in machine learning can be broadly classified into three types: supervised learning, unsupervised learning, and reinforcement learning.

[0140] The purpose of neural network training is to minimize output errors. It is a process that repeatedly inputs training data into a neural network, calculates the error between the network's output and the target for the training data, and updates the weights of each node by backpropagating the network's error from the output layer to the input layer in a direction that reduces the error.

[0141] Supervised learning uses training data with labeled correct answers, whereas unsupervised learning may not have the correct answers labeled in the training data. For example, in the case of supervised learning regarding data classification, the training data may consist of data where each training data point is labeled with a category. Labeled training data is input into a neural network, and an error can be calculated by comparing the network's output (e.g., category) with the labels of the training data. The calculated error is backpropagated within the neural network (i.e., from the output layer to the input layer), and this backpropagation can update the connection weight(s) of each node in each layer of the neural network. The amount of change in the connection weights of each updated node can be determined by the learning rate. The neural network's calculation of the input data and the backpropagation of the error can constitute a learning epoch. The learning rate can be applied differently depending on the number of iterations of the neural network's learning epoch. For example, in the early stages of training a neural network, a high learning rate can be used to allow the neural network to quickly achieve a certain level of performance, thereby increasing efficiency, and in the later stages of training, a low learning rate can be used to increase accuracy.

[0142] The learning method may vary depending on the characteristics of the data. For example, if the goal is to accurately predict data transmitted from the transmitting end at the receiving end in a communication system, it is desirable to perform learning using supervised learning rather than unsupervised learning or reinforcement learning.

[0143] A learning model corresponds to the human brain, and one can consider the most basic linear model. The machine learning paradigm that uses highly complex neural network structures, such as artificial neural networks, as learning models is called deep learning.

[0144] The neural network cores used for learning methods mainly include deep neural networks (DNN), convolutional neural networks (CNN), and recurrent neural networks (RNN).

[0145] Figure 7 illustrates a perceptron structure used in an artificial neural network.

[0146] An artificial neural network can be implemented by connecting multiple perceptrons. Referring to Fig. 7, the input vector x = (x1, x2, ..., x d If ) is input, weight w=(w1,w2, ...,w d The entire process of multiplying by ) and summing all the results, and then applying the activation function σ(·), is called a perceptron. In a large artificial neural network structure, the simplified perceptron structure exemplified in Fig. 7 can be extended and applied to multi-dimensional perceptrons with different input vectors.

[0147] Figure 8 illustrates a multilayer perceptron structure.

[0148] The perceptron structure exemplified in FIG. 7 can be extended into a multilayer perceptron structure having a total of three layers based on input and output values. An artificial neural network having H (d+1) dimensional perceptrons between the first and second layers and K (H+1) dimensional perceptrons between the second and third layers can be represented by the multilayer perceptron structure exemplified in FIG. 8.

[0149] The layer where the input vector is located is called the input layer, the layer where the final output value(s) are located is called the output layer, and all layers located between the input layer and the output layer are called hidden layers. Although the example in Fig. 8 includes three layers, the input layer is excluded when counting the actual number of layers in an artificial neural network; therefore, the artificial neural network based on the multilayer perceptron structure of Fig. 8 can be viewed as consisting of two layers. An artificial neural network is constructed by connecting the perceptrons of the basic blocks in a two-dimensional manner.

[0150] In a neural network, a layer consists of small individual units called neurons. In a neural network, a neuron receives input from other neurons, performs some processing, and produces an output. The region within the previous layer where each neuron receives input is called the receptive field. Each neuron calculates an output value by applying a specific function to the input values ​​received from the receptive field within the previous layer. The specific function applied to the input values ​​is determined by i) a vector of weights and ii) a bias. In a neural network, learning is performed by iteratively adjusting these biases and weights. The vector of weights and the bias are called filters and represent particular features of the input.

[0151] The aforementioned input layer, hidden layer, and output layer can be applied not only to multilayer perceptrons but also to various artificial neural network structures such as CNNs, which will be described later. As the number of hidden layers increases, the artificial neural network becomes deeper, and a machine learning paradigm that uses a sufficiently deep artificial neural network as a learning model is called deep learning. In addition, the artificial neural network used for deep learning is called a deep neural network (DNN).

[0152] The above multilayer perceptron structure is referred to as a fully-connected neural network. In a fully-connected neural network, there are no connections between neurons located in the same layer, and connections exist only between neurons located in adjacent layers. A DNN possesses a fully-connected neural network structure and is composed of a combination of multiple hidden layers and activation functions; it can be usefully applied to identify correlation characteristics between inputs and outputs. Here, correlation characteristics may refer to the joint probability of the input and output.

[0153] Meanwhile, depending on how multiple perceptrons are connected to each other, various artificial neural network structures different from DNNs can be formed.

[0154] Figure 9 illustrates a convolutional neural network (CNN) structure.

[0155] In a DNN, neurons within a single layer are arranged in a one-dimensional manner. However, referring to Fig. 9, in a CNN, it can be assumed that there are w neurons horizontally and h neurons vertically arranged in a two-dimensional manner. In this case, since a weight is applied for each connection leading from an input neuron to a hidden layer, a total of h × w weights must be considered. Since there are h × w neurons in the input layer, there are a total of h between two adjacent layers. 2 w 2 A number of weights are required.

[0156] Figure 10 illustrates a filter operation in a CNN.

[0157] The CNN exemplified in Fig. 6 has a problem in which the number of weights increases exponentially depending on the number of connections, so instead of considering the connections of all neurons between adjacent layers, it is assumed that there are small filters, and as exemplified in Fig. 10, weighted sum operations and activation function operations are performed on the parts where filters overlap.

[0158] A single filter has weights corresponding to its size, and the weights can be trained to extract and output a specific feature on the image as a factor. In Fig. 10, a 3×3 filter is applied to the top-left 3×3 area of ​​the input layer, and the output value obtained by performing a weighted sum operation and an activation function operation on the corresponding neuron is z 22 Save to.

[0159] The filter above scans the input layer, moves by a fixed interval horizontally and vertically, performs weighted sum operations and activation function operations, and places the output value at the current filter position. This method of operation is similar to the convolutional operation on images in the field of computer vision, so a deep neural network with this structure is called a CNN, and the hidden layer generated as a result of the convolutional operation is called a convolutional layer. In addition, a neural network containing multiple convolutional layers is called a deep convolutional neural network (DCNN).

[0160] In a convolutional layer, the number of weights can be reduced by calculating a weighted sum that includes only the neuron(s) located within the area covered by the current filter. As a result, a single filter can be utilized to focus on features of a local area. Accordingly, CNNs can be effectively applied to image data processing where physical distance in a 2-dimensional area serves as an important criterion for judgment. Meanwhile, in a CNN, multiple filters can be applied immediately before the convolutional layer, and multiple output results can be generated through the convolution operation of each filter.

[0161] A fully connected layer connects every neuron in one layer to every neuron in another layer.

[0162] NN systems have the advantage of being able to solve difficult problems or optimize based on non-linearity. For NN-based wireless communication systems, end-to-end learning has been proposed to simultaneously optimize channel coding, modulation, and filtering at the transmitting end, as well as channel estimation and signal detection algorithms at the receiving end.

[0163] End-to-end communication can expect high performance gains compared to existing communication systems, where individual blocks and the transmitter and receiver are optimized separately, because the transmitter NN and receiver NN are trained jointly. Research is underway to solve problems such as CSI acquisition, pilot signal design, data transmission, and beamforming by applying end-to-end communication.

[0164] However, to maximize the performance of end-to-end communication, the transmitter and receiver must be trained to adapt appropriately to the channel environment. For example, an end-to-end communication system can implement the transmitter, channel, and receiver as NN(s) and jointly optimize the trainable parameters of the transmitter and receiver for a specific channel model. However, unlike a system that considers an NN only at the receiver, training of end-to-end communication (hereinafter referred to as end-to-end learning) involves not only transmitting a training signal from the transmitter to the receiver for weight calculation but also requiring feedback from the receiver to the transmitter, which leads to a problem of significantly increased signaling overhead. To solve this, offline learning or methods to train only the NN at the receiver can be considered. However, offline learning has the disadvantage that end-to-end communication cannot operate adaptively on the channel, and training only the receiver NN is a sub-optimal method because the transmitter cannot be fine-tuned as the transmitter NN cannot be trained.

[0165] Figure 11 illustrates the concept of step-to-step learning with backpropagation. In particular, Figure 11 is a conceptual diagram of the training signal and feedback of step-to-step learning with backpropagation. In Figure 11, ∇ t,pis the gradient for the p-th training signal of the t-th batch, and H t,p is the channel experienced by the p-th training signal of the t-th batch, and X t,p is the p-th training signal of the t-th batch, and L t,p is the loss for the p-th training signal of the t-th batch.

[0166] Referring to FIG. 11, backpropagation is performed from the loss value of the receiver calculated from the training signal x input from the transmitter to the receiver. Therefore, to secure optimal performance of end-to-end communication, research is required on an efficient method to jointly train the transmitter and receiver of end-to-end communication. Below, several implementations of the present specification applying end-to-end learning strategies and transfer learning are described.

[0167] Figure 12 shows an example of a method for calculating gradients in a neural network.

[0168] In AI, the learning of NNs is performed by calculating gradients, which are calculated through backpropagation. Backpropagation, that is, the backward propagation of errors, is an algorithm for supervised learning of artificial neural networks using gradient descent. Backpropagation is calculated through the chain rule, as exemplified in Fig. 12. When a specific training signal is given as input, the gradient value for each layer is obtained from the feedforward value calculated as the corresponding training signal passes through each layer of the NN.

[0169] In step-to-step learning as well, gradients can be calculated through backpropagation. Referring to Fig. 12, w tSince the terms differentiated with respect to include the differential terms at the receiver, backpropagation calculated values ​​must be transmitted from the receiver to the transmitter. This significantly increases signaling overhead in end-to-end learning. In Fig. 12, each differential term expressed by the chain rule is a value related to the part connected by an arrow on the transmit / receive path. Additionally, in Fig. 12, t and k represent the t-th batch and the k-th training symbol, respectively, and w nm Eunw t It is an element of my n-th row and m-th column.

[0170] Although NN systems have various advantages, to maximize the performance of NN systems, NNs must be trained to adapt appropriately to the channel environment. Rapid training affects the performance of communication systems to which NNs are applied.

[0171] Figure 13 illustrates the structure of a GAN. A GAN enables the generation of data with a distribution similar to the actual data distribution.

[0172] Referring to Fig. 13, the GAN includes a Generator (G) and a Discriminator (D). The role of D is to distinguish whether the input is real data. When data x is given as input, the output D(x) of D returns the probability that x is real data. The role of G is to create fake data such that D cannot distinguish whether it is real. For example, G samples a noise / random number vector z from a standard normal distribution and then uses z as input to generate fake data G(z). G(z) is used as input to D, and D(G(z)) returns the probability that G(z) is real data.

[0173] GANs are trained by updating the weights of D and G, respectively. For D, with G fixed, the weights are updated to return a high probability for real data (x–Pdata(x)) and a low probability for similar data (z–Pz(z))). For G, the weights are updated to return a high probability when G(z) is input to the previously trained D. By training G and D alternately in this manner, G can be trained to generate similar data so well that D cannot distinguish it, while D can be trained to distinguish any similar data generated by G.

[0174] Mathematical Equation 1 represents the objective function of a GAN. In the case of a GAN, there are two networks to be trained, and since conflicting training occurs, optimization is performed separately.

[0175] [Mathematical Formula 1]

[0176]

[0177] Equations 2 and 3 represent the objective function of D and the objective function of G, respectively. For D, the weights are updated in the direction of maximizing V(D,G) with respect to D. For G, the weights are updated in the direction of minimizing V(D,G) with respect to G.

[0178] [Mathematical Formula 2]

[0179]

[0180] [Mathematical Formula 3]

[0181]

[0182] For each symbol in mathematical formulas 1 to 3, the following may be referenced.

[0183] - E: Represents the expected value / average value (expectation), and

[0184] - D(xi): Represents the probability that the input xi is actual data (i=1~m),

[0185] - G(zi) represents similar data from the input zi (i=1~m).

[0186] Figure 14 is a diagram illustrating a VAE model.

[0187] A Variational Autoencoder (VAE) can be a generative model composed of an encoder and a decoder. The encoder encodes input data into a latent space, and the decoder can reconstruct the input data by taking a vector from the latent space as input. Here, the encoder and decoder can each be composed of a multilayer perceptron (MLP).

[0188] Specifically, a VAE can be a stochastic autoencoder model that learns the latent distribution of input data. Unlike general autoencoders, a VAE can encode input data to transform it into a distribution in a latent space, and then reconstruct the original data by sampling from this distribution and decoding. Through this process, it is possible to model the complex structure of data and learn the statistical characteristics of the data distribution.

[0189] VAE-based data imputation can replace missing parts of data using values ​​sampled from a latent space. For example, during the encoding phase, a VAE can input data containing missing values ​​into an encoder to map it to the latent space. In the sampling phase, the distribution of the missing data in the latent space can be sampled to generate latent variables that fill those missing values. In the decoding phase, these latent variables are restored to the original data space through a decoder to complete the data containing missing values. Through the repetition of this process, the model can be optimized to provide more precise imputation results by repeatedly predicting and correcting missing values.

[0190] In other words, because VAE-based imputation considers the potential distribution of the data, it can compensate for missing values ​​so that they take on values ​​close to the original distribution. In particular, VAE can perform exceptionally well with high-dimensional data or data with complex distributions.

[0191] The training process of a VAE can be broadly divided into two stages. The first stage may be random sampling in the latent space. In this process, the mean and variance are calculated, and random sampling can be performed from a normal distribution based on the calculated mean and variance. The second stage may be the process of minimizing a loss function using the reconstruction error and KL divergence.

[0192] Such VAEs can be effectively utilized in relation to RF sensing. For example, VAEs can filter noise from data related to RF sensing. In this case, VAEs can be effectively utilized in environments with high interference. Additionally, as mentioned above, VAEs can be used for data impudation. VAEs can reconstruct missing data points in datasets related to RF sensing. Such reconstruction of data points can be essential for the accurate localization of devices and environmental monitoring. For example, VAEs can generate a complete dataset that supplements missing RF signal data by inferring missing RF signal data based on observed or sensed data due to sparse sensor placement. Furthermore, in latent representation, the encoder-decoder architecture of VAEs can capture the underlying patterns of RF signals, enabling robust modeling even under noisy or unstable conditions. For example, VAEs can improve accuracy and enable more precise location measurements by supplementing missing Received Signal Strength (RSS) data in indoor location scenarios.

[0193] Meanwhile, regarding noise removal from RF signals, a Diffusion Model (DM) may be additionally considered. Diffusion models or diffusion-based noise reduction modules can effectively remove noise from RF signal data by utilizing the diffusion model. For example, a diffusion model can effectively perform noise removal from data by learning a process of progressively contaminating the original data with noise and then restoring the noise-added data to a clean state.

[0194] Specifically, a diffusion model is a deep learning model that contaminates the original data by gradually adding noise and then restores it in reverse to remove the noise or generate the data. The basic concept lies in regenerating the original data through a "backdiffusion" process that starts with the input data filled with noise and gradually restores it to the original data.

[0195] The operating principle of the diffusion-based noise reduction module is as follows.

[0196] - Noise Addition Step (Forward Process): A step that modifies the original data by adding noise to it.

[0197] - Noise Reduction Learning (Reverse Process): The stage of learning the process of restoring noisy data to its clean original state. Through this, the model can learn how to remove noise from the input data.

[0198] - Noise Removal (Inference Process): When new data containing noise is input through a trained model, the model reconstructs the data in reverse, removing the noise and outputting data close to the original.

[0199] Such a spreading model can effectively remove complex noise contained in RF signals. In other words, by learning complex patterns associated with RF signals and inferring in reverse based on the learned patterns, the spreading model can effectively remove even complex forms of noise.

[0200] More specifically, diffusion models can transform sparse data into a high-resolution representation by generating missing details. Such transformation can be essential in environments with low sensor density, such as remote areas or large-scale IoT facilities. Alternatively, diffusion models can generate robust RF signal data corresponding to the RF signal data. Through noise reduction mechanisms, diffusion models can generate high-quality synthetic RF data that can aid in testing IoT applications in various scenarios. Furthermore, diffusion models can improve IoT security applications by modeling the precise distribution of input data to accurately detect deviations. For example, by filling in details to improve sparse RF maps, diffusion models can effectively support more accurate environmental monitoring or obstacle detection in industrial IoT applications.

[0201] Below, ISAC is explained in detail.

[0202] FIG. 15 illustrates examples of wireless sensing modes supported by ISAC.

[0203] Referring to Fig. 15, when considering the transmission and reception operations in the 3GPP standard and the nodes participating therein, the sensing mode can be broadly classified as follows.

[0204] (a) BS mono-static sensing mode: The BS that transmitted the radio wave receives the reflected signal.

[0205] (b) BS-to-BS bi-static sensing mode: Another BS receives the reflected signal of a radio wave transmitted by a specific BS.

[0206] (c) BS-to-UE bi-static sensing mode: The UE receives the signal reflected from the radio wave transmitted by the BS.

[0207] (d) BS mono-static sensing mode: The UE that transmitted the radio wave receives the reflected signal

[0208] (e) UE-to-UE bi-static sensing mode: Another UE receives the reflected signal of a radio wave transmitted by a specific transmitting UE

[0209] (f) UE-to-BS bi-static sensing mode: The BS receives the reflected signal of the radio wave transmitted by the transmitting UE.

[0210] However, in addition to the six use cases mentioned above, a sensing mode including multiple transmitting / receiving nodes may be referred to using the term multi-static sensing mode.

[0211] Wireless sensing via ISAC / JCAS is being considered for application in various scenarios. Generally, wireless sensing is considered for the purpose of acquiring information about targets that do not have a communication module (or are independent of a communication module), and for example, the scenarios that can be considered can be broadly classified into three types.

[0212] (1) Object detection and tracking: This is a scenario for sensing target objects or people or tracking location information. Representative scenarios include intruder sensing in indoor / outdoor situations, location tracking of UAVs or AGVs, and support for autonomous driving.

[0213] (2) Environment monitoring: This is a scenario intended to collect information about the environment around the transmitting / receiving nodes. Examples of scenarios to consider include rainfall information observation and flood sensing.

[0214] (3) Motion monitoring: This is a scenario for sensing the motion of a target, and typical examples include scenarios for distinguishing human movements or gestures.

[0215] The performance metrics and levels required for each of the above scenarios are diverse and may differ from one another. To design an ISAC / JCAS suitable for the quality of service required in each scenario, various key performance requirements need to be considered. The 3GPP standard TS 22.137 document defines the positioning estimation accuracy, velocity estimation accuracy, confidence level, sensing resolution, missed detection probability, false alarm probability, maximum latency of the sensing service, and refreshing rate as key performance requirements for each service scenario, and the required levels for each key performance requirement may vary depending on the service scenario.

[0216] Radio frequency sensing capabilities can provide services for determining object locations without devices because they do not require connecting to objects via devices within a network. The ability to obtain range, velocity, and angle information from radio frequency signals can provide a wide range of new functions, such as various object sensing, object recognition (e.g., vehicles, humans, animals, UAVs), and high-precision localization, tracking, and activity recognition. Radio sensing services can provide information to various industries (e.g., unmanned aerial vehicles, smart homes, V2X, factories, railways, public safety, etc.) that enable applications such as intruder sensing, assisted vehicle steering and navigation, trajectory tracking, collision avoidance, traffic management, and health and traffic management. In some cases, radio sensing may utilize non-3GPP type sensors (e.g., radar, cameras) to further support 3GPP-based sensing. For example, the operation of a radio sensing service—that is, the sensing operation—may rely on the transmission, reflection, and scattering processing of radio sensing signals. Therefore, wireless sensing can provide an opportunity to enhance existing communication systems from communication networks to wireless communication and sensing networks.

[0217] FIGS. 16 and 17 illustrate an example of an application of ISAC to a 3GPP radio communication system. The embodiments of FIGS. 16 and 17 may be combined with various embodiments of the present disclosure. Specifically, FIG. 16 illustrates an example of sensing using a sensing receiver and a sensing transmitter located at the same position (e.g., monostatic sensing), and FIG. 17 illustrates an example of sensing using a separated sensing receiver and a sensing transmitter (e.g., bistatic sensing).

[0218] For example, in the sensing process of ISAC, information about the surrounding environment can be collected by analyzing how the transmitted signal is reflected, scattered, or diffracted. In the case of signal transmission, a transmitter may transmit or radiate a signal for ISAC. As described above, the signal for ISAC may include data and RS. The signal for ISAC may be received by a receiver of a receiving device (e.g., a terminal or a base station). At this time, the receiver may receive the signal for ISAC and the reflected wave of the signal. In this case, the receiving device may simultaneously analyze the received signal and the reflected signal.

[0219] For example, the receiving device can decode the data through a direct path signal (LoS, Line of Sight) and extract the distance, speed, and direction of the object through the reflected signal or the reflected path signal (NLoS, Non-Line of Sight). For example, the receiving device can estimate the distance to the (surrounding) object and the speed of the object by applying the FMCW (Frequency-Modulated Continuous Wave) technique to the reflected signal, or detect the relative speed to the (surrounding) object by applying the Doppler Shift Analysis technique.

[0220] Below, we will explain in detail how to perform ISAC more effectively and accurately using the neural network model described above.

[0221] Generative AI System for Enhancing Sparse Radar Sensing in ISAC Systems

[0222] ISAC systems play a crucial role in fields such as autonomous vehicles, smart city infrastructure, and industrial monitoring, where radar sensing is essential for interpreting environmental conditions, detecting objects, and enabling safe navigation. However, if the ISAC system is applied to a low-frequency band (e.g., 15 GHz or higher) lower than the frequency band in which conventional radar operates (e.g., 77 GHz or higher), the radar data acquired through the ISAC may have lower resolution compared to conventional radar. In this case, the performance of the ISAC based on the radar data may be limited. For instance, when the ISAC operates at a low frequency lower than that of conventional radar, the bandwidth may also be narrower than that of conventional radar. In this case, the resolution of the radar data acquired by the ISAC may be significantly degraded due to limitations in wavelength and bandwidth. Here, low-resolution radar data can also be referred to as sparse radar data.

[0223] As such, ISAC systems operating under low-frequency bandwidth constraints may experience a significant deterioration in their ability to perform critical functions, such as object detection and accurate localization, due to the acquisition of sparse radar data. In this case, the limited density and resolution of radar data can hinder system efficiency in complex environments, such as urban or industrial settings.

[0224] As such, when there are constraints on frequency and bandwidth, the accuracy and efficiency of sensing operations via ISAC can be significantly degraded. Therefore, a method may be required to ensure that the accuracy and efficiency of sensing are not compromised, even when ISAC is performed in the low-frequency band.

[0225] Therefore, the proposed invention proposes a method to compensate for the degradation of resolution by additionally utilizing generative AI. For example, the proposed invention can acquire high-resolution radar data corresponding to low-resolution radar data by utilizing generative AI. Specifically, the generative AI available may include the aforementioned Variational Autoencoders (VAE), Diffusion Models (DM), and Generative Adversarial Networks (GAN). Such generative AI can provide a solution capable of performing robust sensing by reconstructing radar data acquired from RF signals, removing noise, and augmenting the data. In other words, the proposed invention relates to the use of generative AI to improve sparse radar detection within an Integrated Sensing and Communication (ISAC) system operating under low-frequency bandwidth limitations. This may be intended to optimize radar performance and detection resolution despite limited bandwidth resources by combining enhanced AI-based data imputation and synthetic data generation methods, and to bridge the gap between location, object detection, and environment recognition.

[0226] For example, the proposed invention can overcome the limitations of sparse radar detection in ISAC systems under low-frequency bandwidth conditions by utilizing a generative AI-assisted system. The proposed invention may utilize VAEs for data imputation, DMs for noise removal, and GANs for the generation of high-resolution synthetic data in the high-frequency band. Through the use of such generative AI, the proposed invention can provide a scalable, adaptive, and robust radar detection solution even under low-frequency bandwidth conditions. For example, the proposed invention can design a high-level generative AI system for sparse radar sensing in ISACs under low-frequency bandwidth constraints. This high-level design outlines a generative AI-based system for enhancing the radar sensing capabilities of ISAC systems, particularly in low-frequency bandwidth environments where radar data is scarce. This design can provide a complete pipeline for data imputation, refinement, and synthetic data generation by utilizing three key AI modules: Variational Autoencoders (VAE), Diffusion Models (DM), and Generative Adversarial Networks (GAN).

[0227] Alternatively, the proposed invention can provide a generative AI system for sparse radar detection in an ISAC under low-frequency bandwidth constraints as a Low-Level Design. The Low-Level Design describes specific processing steps, component interconnections, and data flow within each module of the generative AI support system for enhancing radar detection under limited bandwidth conditions. Such a Low-Level Design can be assumed for deployment within an ISAC environment with specific requirements for data imputation, noise reduction, and synthetic data generation.

[0228] In the following, we propose a framework that utilizes generative AI to enhance sparse radar data (e.g., low-resolution radar data) acquired from an ISAC system in the low-frequency band or low-frequency bandwidth. The framework may include the following.

[0229] 1. Data Imputation Module Based on VAE

[0230] The VAE module analyzes and learns available radar patterns to reconstruct and fill in missing radar data points in low-resolution radar data. For example, the imputation module can reconstruct radar data for a limited bandwidth into radar data for a wider bandwidth.

[0231] As described above, the VAE module can fill the gaps in the missing radar data by inserting missing radar data points and generating a complete radar frame from the input of sparse radar data. The VAE module receives sparse radar data (or sparse radar frames) from a data collector (or RF receiver) and outputs densed radar data (or, i.e., imputed radar data) inferred from the missing data points in the sparse radar data. Here, the output radar data can be input to a DM-based noise reduction module to be described later.

[0232] Specifically, in the encoder stage, the input layer of the VAE module can take sparse radar data as a 2D input tensor. The VAE module can apply multiple convolution layers configured with filters to the 2D input tensor to capture spatial dependencies of the radar data and encode sparse features. The VAE module can map the encoded radar data (or sparse features) into a latent space. At this time, each radar data (or radar frame) can be transformed into a probability distribution with mean and variance parameters.

[0233] In the decoder stage, the VAE module can reconstruct entire radar data or radar frames by sampling data points from a latent distribution through latent sampling. The VAE module can reconstruct dense radar data (or radar frames) from the sampled latent vectors through convolutional decoding and generate predictions for missing data. The output layer of the VAE module can output complete radar data (or radar frames) containing estimates for sparse regions for noise reduction. Here, the intrinsic parameters of the VAE module may include latent dimension size and reconstruction loss. Here, the latent dimension size may be a parameter that controls complexity to allow the model to balance detail and computational efficiency. Reconstruction loss may be a parameter for measuring the difference between the original sparse input and the reconstructed output to optimize the accuracy of the imputation.

[0234] To this end, a VAE-based data imputation module (hereinafter referred to as the VAE module) can be trained to identify patterns and spatial relationships in sparse radar data / signals and to predict / infer and fill in missing data points (e.g., data points for adjacent frequency regions outside the existing bandwidth). The VAE module can generate a continuous latent representation of the radar data to accurately fill in missing data points or data gaps. The VAE module can output radar data in which the missing data points are filled in or imputed (e.g., radar data for a second bandwidth wider than the first bandwidth of the input radar data).

[0235] 2. Noise reduction module based on a diffusion model (DM) (hereinafter, DM module)

[0236] The DM module iteratively refines imputed radar data to ensure clarity and robustness even in bandwidth-limited scenarios.

[0237] The DM module receives the imputed radar data and can enhance the clarity and structure of the radar data by repeatedly removing noise from the input radar data. The DM module takes radar data imputed by the VAE module (e.g., high-density radar data / frames) as input and can output refined radar data with minimized noise and improved detail. Specifically, the DM module can perform forward diffusion. Forward diffusion may be a process of gradually applying Gaussian noise to the imputed radar data to progressively transform it into a noise distribution for the imputed radar data over several stages. At this time, the DM module can prevent excessive distortion by maintaining the core structure through noise scheduling to dynamically adjust the noise level. Next, the DM model can perform reverse diffusion. Reverse diffusion may include a noise removal step and a final reconstruction step. The noise removal step may be a step that gradually removes or reverses noise from high-noise radar data using forward spreading to convert it into the state of initial radar data (e.g., imputation data input from a VAE module). In this case, the high-noise radar data can be refined and converted into more consistent and detailed radar data. The final reconstruction step may be a step that reconstructs high-resolution radar data with noise reduced to a negligible level suitable for ISAC operations and outputs the reconstructed high-resolution radar data. Here, internal parameters may include the number of spreading steps and noise dispersion control. The number of spreading steps is a parameter for determining the granularity of the noise removal process; the more steps there are, the finer the details (or detailed features in the radar data) can become.Noise variance control can be a parameter for dynamically adjusting noise variance for optimized data cleaning.

[0238] For example, the DM module can improve the clarity of the output data (or output signal) of the VAE module and enhance its robustness (or resilience) against interference by refining the output data through a method of repeatedly applying and removing noise from the output data of the VAE module. Through the noise removal process described above, the diffusion model can output refined radar data (hereinafter, refined radar data) suitable for applications requiring high accuracy, such as environment sensing and object detection in autonomous vehicles.

[0239] 3. GAN-based synthetic data generation module (hereinafter, GAN module)

[0240] A GAN module can generally generate high-fidelity synthetic radar data that mimics conditions observed in higher bandwidth environments. In this case, the GAN module can improve the synthesis capability and accuracy of the synthetic radar data by utilizing training data that can improve actual performance.

[0241] For example, the GAN module can generate high-resolution synthetic radar data that mimics high-frequency band conditions using refined radar data acquired from the DM module. Such high-resolution synthetic radar data extends the input refined radar data to data for higher frequency bands, and through the synthetic radar data, enables more accurate machine learning model training and robust real-world application performance.

[0242] Specifically, the GAN module can generate high-fidelity synthetic radar data to simulate conditions in high-frequency bandwidths. The synthetic radar data generated in this way can be used to expand original radar data or radar datasets and to improve the training of neural network models based on radar data. The GAN module can receive radar data / frames refined in a DM model as input and output high-resolution synthetic radar data / frames. The high-resolution synthetic radar data / frames can be stored in a data fusion and storage module. Here, the GAN module may include a generator (or Generator Network) and a discriminator (or Discriminator Network).

[0243] The generator may include an input layer, deconvolution layers, and an output layer. The input layer may receive latent noise vectors and conditioned, refined radar data from a DM module. The deconvolution layer may upscale the noise vectors to generate high-resolution radar data / frames that approximate high-bandwidth radar data (or refined radar data). The output layer may generate and output synthetic radar data / frames in which detailed information matching actual high-frequency radar data / frames is added to the high-resolution radar data. For example, the high-resolution radar data upscaled by the deconvolution layer may be radar data for a first band. In this case, the output layer may generate / output synthetic radar data for a second band by adding actual characteristics for a second band, which is a higher frequency band than the first band (e.g., high-frequency components in the second band), to the high-resolution radar data.

[0244] The discriminator may include an input layer, a convolution layer, and a loss calculation unit. The input layer may receive both actual radar data (e.g., refined radar data) and synthetic radar data (radar data output by the generator) for authenticity evaluation. The convolution layer can distinguish between synthetic radar data and actual radar data by analyzing the structural and textural features of the radar data. The loss calculation unit calculates adversarial loss by comparing the generated synthetic radar data with actual radar data (e.g., high-bandwidth radar data or refined radar data) and iteratively optimizes the generator and the discriminator. For example, the discriminator can determine whether the synthetic radar data possesses the desired actual characteristics. For example, if the desired actual characteristics are characteristics of a second band in the high-frequency band, the discriminator can determine whether the synthetic radar data possesses the actual high-frequency characteristics for the second band.

[0245] The internal parameters of the above GAN module may include Adversarial Loss and a Learning Rate. The Adversarial Loss may be a parameter for achieving an optimized balance between the generator and the discriminator so that actual high-frequency band radar data and synthetic radar data cannot be distinguished. The Learning Rate may be a parameter for balancing quality and computational efficiency by controlling the convergence speed of the GAN model.

[0246] For example, the GAN module can be trained by adjusting the parameters of the generator and / or discriminator so that adversarial loss according to the discriminator is minimized. The GAN module trained in this way can generate synthetic radar data for a second band, which is a higher frequency band than the first band, using radar data for the first band.

[0247] In the following, we will explain in detail a method for converting sparse radar data acquired from an ISAC system constrained to a low-frequency band into synthetic radar data for a high-resolution high-frequency band using the aforementioned generative AI model, and for effectively and accurately sensing the surrounding environment based on the said synthetic radar data.

[0248] FIG. 18 is a diagram illustrating how a device performs ISAC using an AI model.

[0249] Referring to FIG. 18, the device can collect radar data through signals for data transmission and sensing in relation to ISAC (S181). For example, the device can collect sparse radar data under band-limited conditions through a Radar Sensing Data Collector. As described above, the bandwidth may be limited to a band for frequencies lower than the band used by conventional radar. In this case, ISAC may collect radar data restrictively for frequencies lower than those used by conventional radar and for a bandwidth narrower than the bandwidth used by conventional radar. For example, ISAC may be performed based on signals for a low-frequency band adjacent to 15 GHz, which is lower than 67 GHz of conventional radar.

[0250] Specifically, a radar sensing data collector can collect low-frequency and bandwidth-limited radar signals / radar data (e.g., sparse radar data, raw radar data) that generally generate sparse and low-resolution radar data / frames. For instance, a radar sensing data collector can receive / acquire environmental radar echoes (e.g., reflected signals of signals for ISAC) under bandwidth constraints. The radar sensing data collector can digitize radar signals / data / frames at preset time intervals / periods and prepare them for processing by a generative AI. Alternatively, the radar sensing data collector can ensure real-time processing by the generative AI by compressing or prioritizing essential data segments from the sparse radar data to minimize latency in the generative AI.

[0251] As such, radar sensing data collectors can collect sparse radar data in frequency and bandwidth-constrained ISAC environments. Radar data may typically include incomplete or low-resolution radar frames affected by low-frequency constraints. Radar sensing data collectors can serve as the initial input for pipelines related to Generative AI.

[0252] In this way, the collected radar data (hereinafter referred to as sparse radar data) can be transmitted from the radar sensing data collector to the VAE module.

[0253] Next, the device can perform imputation on sparse radar data using a VAE module (S183). The sparse radar data can be transmitted to a VAE module that imputes lost / missing radar data points. The VAE module can predict lost / missing radar data points (or values ​​in the sparse region) and output a complete radar frame. For example, the sparse radar data may include frequency information for a first bandwidth, and the VAE module may imputate the sparse radar data into radar data for a second bandwidth (or high bandwidth) that is wider than the first bandwidth. In this case, based on the sparse radar data, the VAE module can supplement the missing / lost data points for the remaining bandwidth excluding the first bandwidth from the second bandwidth, thereby expanding the sparse radar data of the first bandwidth into high bandwidth radar data that has been expanded to the second bandwidth. For example, the VAE module can generate radar data with a higher bandwidth by compressing and representing the characteristics of sparse radar data in a latent space through an encoder, and by generating new samples within the latent space based on the characteristics of the compressed data.

[0254] As described above, the VAE can generate a continuous and dense data representation by filling in lost or missing data points in sparse radar data / frames. Specifically, when the sparse radar data is input, the VAE's encoder can capture spatial and temporal dependencies by compressing the sparse radar data into a latent representation. The VAE's decoder can reconstruct high-bandwidth radar data (or a complete radar frame) by predicting and supplementing the missing data points (e.g., data points for frequency bands outside the bandwidth of the sparse radar data). The VAE can output densed radar data (or high-bandwidth radar data) containing estimated values ​​for the missing data points (or regions).

[0255] In this way, the VAE module can generate denser, high-bandwidth radar data (or data sets) for further refinement according to the DM module.

[0256] Next, the device can perform noise removal and refinement on high-bandwidth radar data using a DM module (S185). Specifically, the DM module can receive radar data generated by expanding to high bandwidth from the VAE module, and can generate / output high-quality radar data by repeatedly performing noise removal and refinement on the input radar data. The DM module can transmit the output radar data to a GAN module for synthesis and augmentation.

[0257] As described above, the DM module can apply noise to the output radar data of the VAE through forward diffusion and estimate / calculate the distribution / variance of the noise. The DM module can repeatedly remove the applied noise through reverse diffusion and enhance the quality and fidelity of the data.

[0258] Next, the device can use a GAN module to generate synthetic radar data (or, high-fidelity radar data / frames) that mimic data conditions for high frequency (and / or high bandwidth) based on the noise-removed and refined radar data (S187). For example, as described above, the device can use the GAN module to generate synthetic radar data for a second frequency band by applying characteristics for a second frequency band higher than the first frequency band to radar data for a first frequency band.

[0259] As described above, the GAN module can generate high-resolution synthetic radar data / frames that mimic high-bandwidth and / or high-frequency bands, and improve the safety of the ISAC system by expanding the training data using said synthetic radar data / frames. Specifically, the GAN module may receive radar data refined by the DM module as input. The generator of the GAN module generates high-fidelity data / frame samples using said refined radar data, and the discriminator of the GAN module can determine whether said generated samples (or synthetic radar data) are similar to or correspond to actual radar data. In this way, the GAN module can generate said synthetic radar data by simulating the characteristics / conditions of radar data in high-bandwidth / high-frequency bands and adding realism and detail to said refined radar data. In this manner, the GAN module can generate synthetic radar data / frames that improve the training and operational accuracy of the AI ​​model.

[0260] Next, the device can train an AI model based on the synthetic radar data or perform operations related to ISAC (S189). For example, the device can train a machine learning model by merging actual radar data (radar data acquired by a receiver) with synthetic radar data, or form / configure a comprehensive data set that can be directly used in real-time applications in an ISAC system (scenario). Alternatively, the device can calculate mobility information regarding surrounding objects related to ISAC based on the synthetic radar data, or transmit the synthetic radar data to another device so that such mobility information can be calculated.

[0261] Alternatively, the device may perform data fusion / storage. For example, the device may collect the outputs of all the modules described above and integrate / fuse them into comprehensive radar data sets. The device may store both actual radar data (or sparse radar data or refined radar data) and synthetic radar data to train machine learning models or use them directly in ISAC applications.

[0262] Alternatively, the device may include a data fusion and storage module. The data fusion and storage module may integrate real radar data (or sparse radar data or refined radar data) and synthetic radar data to form a dataset and support access to said dataset for machine learning training or real-time use in an ISAC system. Here, the data fusion and storage module may receive real radar data / frames and synthetic radar data / frames as input from a GAN module and output an integrated high-resolution / high-quality radar dataset usable in downstream ISAC applications. Specifically, the data fusion and storage module may combine real radar data (or sparse radar data or refined radar data) and synthetic data generated by the GAN into a structured storage format (e.g., a multidimensional array). The data fusion and storage module may label the synthetic radar data and real radar data for the purpose of training the generative AI and support flexible access based on data type, time, and environmental context (data labeling and indexing). The data fusion and storage module can improve the data access speed of real-time applications by applying lossless compression to optimize storage of synthetic radar data and real radar data without loss of quality.

[0263] In summary, the device collects sparse radar data / frames through a radar sensing data collector and transmits the sparse radar data / frames to a VAE module for imputation (① Data Collection). The device can imputate missing / lost data points using the VAE module to output high-density radar data / frames (or high-bandwidth radar data / frames) (② Data Imputation). The device can output high-resolution radar data by refining the high-density radar data through noise removal using a DM module (③ Noise Reduction). The device can generate synthetic radar data / frames (or high-fidelity radar frames) that satisfy / simulate high-frequency bandwidth radar conditions using a GAN module (④ Synthetic Generation). The device can combine, label, and store actual radar data and synthetic radar data for use in an ISAC system or for future training of an ML model (⑤ Data Fusion and Storage).

[0264] In addition, real-time processing and feedback mechanisms can be defined for the proposed invention. For example, each module can be designed for sequential processing that enables near-real-time data enhancement essential for applications such as autonomous driving (Real-Time Data Processing). Furthermore, feedback from the GAN discriminator can be used for fine-tuning the GAN generator to generate high-quality synthetic radar data (GAN Discriminator Feedback); or, key parameters such as noise variance, diffusion steps, and adversarial loss can be dynamically adjusted according to system performance (Self-Optimizing Parameters).

[0265] The following technical details may be considered to perform operations related to the proposed invention.

[0266] - Hardware Requirements: High-performance GPUs (Graphics Processing Units) or TPUs (Tensor Processing Units) may be recommended for efficient real-time imputation, noise reduction, and synthetic data generation (GPU / TPU for Real-Time Processing). Lightweight model variants may be required to reduce computing overhead in edge scenarios for mobile applications (Edge Processing for Mobile Applications).

[0267] - Optimization of Generative AI Models: Regarding model lightweighting, model pruning, quantization, or distillation techniques may be considered to optimize models for low-latency processing on resource-constrained devices (Model Pruning and Compression). Adaptive inference mechanisms can support a balance between the energy consumption and performance requirements of Generative AI models (Energy-Efficient Inference).

[0268] Below, the radar data processing pipeline is described in detail in a scenario involving handshake messages between devices (e.g., between a base station and a terminal).

[0269] Figure 19 is a diagram illustrating how a base station and a terminal process radar data.

[0270] Referring to FIG. 19, the base station and the terminal can exchange radar data with each other. Here, the base station may include the VAE module, DM module, and GAN module described above. Alternatively, the terminal may also include the VAE module, DM module, and GAN module described above.

[0271] 1. Initial data collection request (S191)

[0272] The base station may request the UE to transmit sparse radar data for initial collection under limited bandwidth conditions. In this case, the UE may transmit the collected sparse radar data to the base station in response to the base station's request. The base station may transmit the terminal's sparse radar data to a VAE module for imputation.

[0273] 2. Data Imputation Request (S193)

[0274] A base station can input the sparse radar data into a VAE module to obtain high-density radar data (or high-bandwidth radar data) for the sparse radar data. Alternatively, the base station can request imputation of the sparse radar data from the VAE module to obtain high-bandwidth radar data for the sparse radar data from the VAE module. For example, the sparse radar data for a first bandwidth can be reconstructed into high-bandwidth radar data for a second bandwidth wider than the first bandwidth through imputation by the VAE module.

[0275] 3. Noise removal and data cleaning (S195)

[0276] The base station can input the high-bandwidth radar data (or VAE-imputed radar data) into the DM module to obtain refined radar data with noise reduction, smoothing, and resolution enhancement for the high-bandwidth radar data. Alternatively, the base station can obtain refined radar data from the DM module by requesting noise removal, smoothing, and resolution enhancement for the high-bandwidth radar data from the DM module.

[0277] 4. Generation of synthetic radar data (S197)

[0278] A base station may input the refined radar data into a GAN module to obtain synthetic radar data (or synthetic high-fidelity radar data / frames) that mimics the bandwidth for high frequencies of the refined radar data. Alternatively, the base station may request the GAN module to generate synthetic radar data that mimics the bandwidth for high frequencies of the refined radar data. For example, the base station may input the refined radar data having a second bandwidth for a first band into the GAN module to obtain synthetic radar data having characteristics in a second band (or mimicking the characteristics of the second band) which is a high frequency band (e.g., 67 GHz or higher) higher than the first band (e.g., 16 GHz or higher and / or less than 67 GHz).

[0279] 5. Data Fusion and Storage (S199)

[0280] The base station may synthesize / merge actual radar data (e.g., collected radar data) with the synthetic radar data and transmit the merged radar data to the UE as a comprehensive dataset that can be used for ML model training or real-time ISAC applications. Additionally, the base station may notify the UE that operational support related to ISAC is ready. The UE may send an ACK to the base station for the reception of the merged radar data and perform training (training of an AI model for ISAC) based on the merged radar data or perform real-time deployment in an ISAC scenario.

[0281] These sequences (1–5) represent a typical data processing pipeline between a base station and a UE, and each processing step can be captured via module-based communication.

[0282] The high-level system flow is as follows.

[0283] 1. Data flow between modules

[0284] Each module can receive input radar data from the previous module and transmit its own output, processed from the input radar data, to the next module. This ensures a seamless transition from sparse data collection to the generation of high-quality synthetic data.

[0285] 2. Feedback and Optimization Loop

[0286] The GAN module may include an internal feedback loop between the generator and the discriminator, and the quality of synthetic radar data can be improved through iterations via the feedback loop. By integrating an optional feedback loop between the DM module and the VAE module, the accuracy of data imputation and data cleaning can be continuously improved.

[0287] 3. User Access and Real-time Data Usage

[0288] The finally stored radar data (collected radar data, synthesized radar data, and / or merged radar data) can be accessed to train machine learning models or used / accessed in real-time for ISAC tasks such as object detection and localization.

[0289] Alternatively, the following matters may be additionally considered in relation to the proposed invention.

[0290] - Real-time Adaptability: By implementing edge processing, real-time imputation, noise removal, and synthetic data generation can be performed directly at the data collection point, making it useful for autonomous systems.

[0291] - Automated Model Retraining: By integrating feedback from deployed systems, the generative AI model can be continuously retrained to optimize performance as new radar data is collected.

[0292] - Energy Efficiency: By using variations of the lightweight model for edge deployment, system resource efficiency can be increased, particularly for mobile or remote ISAC applications.

[0293] Here, a Generative AI-enabled system can handle the problem of sparse radar data sensed in an ISAC system environment with limited low-frequency bandwidth. Specifically, by sequentially utilizing VAE module(s), DM module(s), and GAN module(s), the system can provide high-quality data that is difficult to achieve under limited bandwidth conditions. Through this, the system can provide radar data that is highly useful for complex IoT, automotive, and industrial monitoring applications.

[0294] FIG. 20 is a diagram illustrating a method for a terminal to receive second radar data generated based on first radar data from a base station.

[0295] Referring to FIG. 20, the terminal can transmit a signal for data transmission and sensing to a base station (S201). As described above, the signal is a signal for ISAC and may be a signal intended for sensing the surrounding environment as well as data transmission.

[0296] Next, the terminal receives a reflected signal of the signal to acquire first radar data for a first frequency band and can transmit the first radar data to the base station (S203). Here, the first radar data may be defined as first reflected data. For example, the terminal may acquire the first radar data as sparse radar data (or low-resolution radar data) due to frequency band limitations in an ISAC communication environment. Here, the terminal may transmit the first radar data to the base station to acquire synthetic radar data that is synthesized to reflect characteristics of the high frequency band and compensates for missing values, refines, and is not immediately performed on the surrounding environment based on the first radar data.

[0297] Next, the terminal can receive second radar data from the base station, which includes synthetic radar data generated based on the first radar data (S205). Here, the synthetic radar data is synthetic radar data synthesized using the first radar data to have characteristics for the second frequency band, as described in the definition of "Generative AI System for Enhancing Sparse Radar Sensing in ISAC Systems".

[0298] Here, the second frequency band is a high frequency band higher than the first frequency band and may have a frequency bandwidth wider than the first frequency bandwidth. For example, the second frequency band may have a starting frequency that is higher by a preset frequency than the end frequency of the first frequency band and may have a frequency bandwidth wider than the bandwidth of the first frequency band.

[0299] For example, as described in the definition of "Generative AI System for Enhancing Sparse Radar Sensing in ISAC Systems," the synthetic radar data may include synthetic radar data generated through sequential processing of the first radar data by a first model based on a Variational Autoencoder (VAE), a second model based on a Diffusion Model (DM), and a third model based on Generative Adversarial Networks (GAN). Specifically, the first VAE-based model can generate supplementary radar data for a second bandwidth wider than the first bandwidth of a first frequency band by supplementing missing values ​​or radar values ​​based on the first radar data. For example, the first VAE-based model can predict missing radar values ​​in a second bandwidth wider than the first bandwidth based on radar values ​​acquired in the first bandwidth of the first frequency band, and output supplementary radar data for the second bandwidth by supplementing the first radar data with the predicted values. For example, the above VAE-based first model can extend the above first radar data for the above first bandwidth into radar data for the above second bandwidth (e.g., complementary radar data). The above DM-based second model receives the complementary radar data, which is the above bandwidth-extended first radar data, and outputs refined radar data by removing noise from the complementary radar data through forward diffusion and reverse diffusion.The above GAN-based third model receives the above refined radar data as input and can output the above synthesized radar data synthesized using the input refined radar data to have frequency characteristics for a second frequency band higher than the first frequency band (e.g., a frequency pattern or harmonic function corresponding to the second frequency band).

[0300] For example, the above VAE-based first model outputs supplementary radar data that expands the bandwidth of the first radar data by predicting / supplementing missing values ​​on the first radar data for a second bandwidth wider than the bandwidth of the first frequency band, the above DM-based second model outputs refined radar data that is refined by removing noise from the supplementary radar data, and the above GAN-based third model outputs the synthetic radar data synthesized to mimic / reflect the frequency characteristics of the second frequency band, which is a high frequency band higher than the first frequency band, while maintaining the original characteristics of the first radar data using the input refined radar data.

[0301] In this case, the terminal can acquire high-resolution synthetic radar data from the base station, which increases the resolution of the first radar data while maintaining the characteristics of the original data of the first radar data as much as possible. At this time, the terminal can perform a sensing operation regarding the surrounding environment related to the ISAC based on the synthetic radar data, or train an artificial neural network model that performs the sensing operation of the ISAC.

[0302] Alternatively, without going through the base station, the terminal can directly use the first model, the second model, and the third model to generate / acquire synthetic radar data corresponding to the first radar data, and can perform an ISAC operation based on the synthetic radar data.

[0303] FIG. 21 is a diagram illustrating a method for a base station to acquire second radar data generated based on first radar data received from a terminal.

[0304] Referring to FIG. 21, the base station may receive first radar data regarding a reflected signal of a signal for data transmission and sensing from a terminal (S211). For example, the base station may receive a signal for data transmission and sensing from the terminal and receive first radar data regarding a reflected signal of the signal collected / received by the terminal. For example, the base station may receive the first radar data from the terminal, which is sparse radar data received in an ISAC communication environment where the frequency band is limited.

[0305] Next, the base station may generate synthetic radar data based on the first radar data (S213). For example, the base station may generate synthetic radar data, which is high-resolution radar data corresponding to the first radar data, using a first model, a second model, and a third model as described in the "Generative AI System for Enhancing Sparse Radar Sensing in ISAC Systems", and may generate the second radar data based on the synthetic radar data. For example, the second frequency band may be a high-frequency band higher than the first frequency band and may have a frequency bandwidth wider than the first frequency band. For example, the second frequency band may have a starting frequency that is higher by a preset frequency than the end frequency of the first frequency band and may have a frequency bandwidth wider than the bandwidth of the first frequency band. For example, the

[0306] For example, as described in the definition of "Generative AI System for Enhancing Sparse Radar Sensing in ISAC Systems," the base station can generate synthetic radar data corresponding to the first radar data through sequential processing of the first radar data using a first model based on a Variational Autoencoder (VAE), a second model based on a Diffusion Model (DM), and a third model based on Generative Adversarial Networks (GAN). Specifically, the base station can input the first radar data into the first model based on the VAE and obtain supplementary radar data from the first model based on the VAE, wherein the first radar data is supplemented to have radar data for a second bandwidth that is wider than the first bandwidth of the first frequency band. For example, as described above, the first model based on the VAE can generate supplementary radar data for a second bandwidth that is wider than the first bandwidth by predicting / supplementing missing values ​​for adjacent bands outside the first bandwidth based on the first radar data. Subsequently, the base station inputs the complementary radar data into the DM-based second model and can obtain refined radar data from the DM-based second model, in which noise has been removed from the complementary radar data. For example, the DM-based second model can output refined radar data in which noise has been removed from the complementary radar data by applying forward diffusion and reverse diffusion to the complementary radar data. Subsequently, the base station inputs the refined radar data into the GAN-based third model to obtain synthetic radar data synthesized to mimic the characteristics (e.g., frequency patterns, harmonic components, etc.) of a second frequency band, which is a higher frequency band than the first frequency band.For example, the GAN-based third model may generate candidate synthetic radar data by adjusting the parameters of the generator using a generator and a discriminator until the candidate synthetic radar data generated by the generator is not identified as fake, and may output the candidate synthetic radar data that is not identified as fake through the discriminator as the synthetic radar data. In this case, the synthetic radar data may include high-resolution radar data with a wider bandwidth and a higher frequency band than the first radar data.

[0307] Next, the base station may transmit to the terminal a second radar data including the synthetic radar data and / or the first radar data (S215). Alternatively, the second radar data may include at least one of the first radar data, supplementary radar data, refined radar data, and synthetic radar data.

[0308] Thus, the proposed invention can effectively expand low-resolution radar data acquired in an ISAC environment limited to a low-frequency band (e.g., 15 GHz or higher) into high-resolution radar data corresponding to the frequency band in which existing radars operate (e.g., 77 GHz or higher) by using a model based on an artificial neural network. Alternatively, the proposed invention can effectively expand the low-resolution radar data into high-resolution radar data satisfying various bandwidth conditions by using a model based on an artificial neural network. Alternatively, the proposed invention can significantly improve the accuracy of an ISAC system even in an ISAC environment limited to a low-frequency band by performing sensing operations, etc., based on the high-resolution radar data expanded from the low-resolution radar data.

[0309] The embodiments described above are combinations of the components and features of the present invention in a specific form. Each component or feature should be considered optional unless otherwise explicitly stated. Each component or feature may be implemented in a form not combined with other components or features. Additionally, it is possible to construct embodiments of the present invention by combining some components and / or features. The order of operations described in the embodiments of the present invention may be changed. Some components or features of one embodiment may be included in another embodiment, or may be replaced with corresponding components or features of another embodiment. It is obvious that embodiments may be constructed by combining claims that do not have an explicit citation relationship in the claims, or that new claims may be included by amendment after filing.

[0310] In this document, embodiments of the present invention are described primarily with a focus on the signal transmission and reception relationship between a terminal and a base station. This transmission and reception relationship is extended in the same or similar manner to signal transmission and reception between a terminal and a relay or between a base station and a relay. Specific operations described in this document as being performed by a base station may, in some cases, be performed by an upper node. That is, it is self-evident that various operations performed for communication with a terminal in a network consisting of multiple network nodes including a base station may be performed by the base station or other network nodes other than the base station. The base station may be replaced by terms such as fixed station, Node B, eNode B (eNB), and access point. Additionally, the terminal may be replaced by terms such as User Equipment (UE), Mobile Station (MS), and Mobile Subscriber Station (MSS).

[0311] Embodiments according to the present invention may be implemented by various means, for example, hardware, firmware, software, or a combination thereof. In the case of implementation by hardware, one embodiment of the present invention may be implemented by one or more ASICs (application specific integrated circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays), processors, controllers, microcontrollers, microprocessors, etc.

[0312] In the case of implementation by firmware or software, an embodiment of the present invention may be implemented in the form of a module, procedure, function, etc., that performs the functions or operations described above. Software code may be stored in a memory unit and executed by a processor. The memory unit may be located inside or outside the processor and may exchange data with the processor by various means already known.

[0313] It is obvious to those skilled in the art that the present invention may be embodied in other specific forms without departing from the features of the invention. Accordingly, the above detailed description should not be interpreted restrictively in all respects and should be considered exemplary. The scope of the invention shall be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention.

[0314] The present disclosure may be used in various devices, including networks such as terminals, base stations, and / or location servers of wireless mobile communication systems.

Claims

1. In a method using a terminal, A step of transmitting a signal for data transmission and sensing to a base station; A step of receiving a reflected signal of the above signal to acquire first radar data for a first frequency band and transmitting the first radar data to the base station; and The method includes the step of receiving second radar data including synthetic radar data generated based on the first radar data from the base station, A method in which the above-mentioned synthetic radar data is radar data synthesized based on the above-mentioned first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

2. In Paragraph 1, A method characterized in that the second frequency band has a starting frequency that is higher than the end frequency of the first frequency band by a preset frequency, and has a frequency bandwidth that is wider than the bandwidth of the first frequency band.

3. In Paragraph 1, A method characterized in that the first radar data is synthesized into the synthetic radar data through sequential processing by a first model based on VAE (Variational Autoencoder), a second model based on DM (Diffusion Model), and a third model based on GAN (Generative Adversarial Networks).

4. In Paragraph 3, A method characterized in that the above-described VAE-based first model is a model that outputs supplementary radar data that supplements missing values ​​for a bandwidth wider than the bandwidth of the first frequency band based on the above-described first radar data.

5. In Paragraph 3, A method characterized in that the DM-based second model is a model that outputs refined radar data by removing noise from the complementary radar data output by the VAE-based first model through input forward diffusion and reverse diffusion.

6. In Paragraph 3, The above GAN-based third model is a model that outputs the synthesized radar data synthesized to have frequency characteristics for a second frequency band higher than the first frequency band based on the above first radar data, and A method characterized in that the second frequency band has a starting frequency that is higher than the end frequency of the first frequency band by a preset frequency and has a frequency bandwidth wider than the bandwidth of the first frequency band.

7. In Paragraph 3, The above VAE-based first model outputs supplementary radar data that supplements missing values ​​for a bandwidth wider than the bandwidth of the first frequency band based on the above first radar data, and The above DM-based second model receives the above complementary radar data as input and outputs refined radar data with noise removed from the above complementary radar data through forward diffusion and reverse diffusion, and A method characterized in that the above-described GAN-based third model receives the above-described refined radar data as input and outputs the above-described synthesized radar data synthesized to mimic the frequency characteristics of a second frequency band higher than the first frequency band based on the above-described refined radar data.

8. In Paragraph 1, A method characterized in that the terminal trains a sensing operation related to ISAC (Integrated Sensing and Communication) or a model based on an artificial neural network related to ISAC based on the second radar data.

9. A computer-readable recording medium storing a program for performing the method described in paragraph 1.

10. RF (Radio Frequency) transceiver; and It includes a processor connected to the above RF transceiver, and The processor controls the RF transceiver to transmit a signal for data transmission and sensing to a base station, receives a reflected signal of the signal to acquire first radar data for a first frequency band, and receives second radar data from the base station including synthetic radar data generated based on the first radar data. The above-mentioned synthetic radar data is radar data synthesized based on the above-mentioned first radar data to reflect the characteristics of a second frequency band higher than the first frequency band, and is a terminal.

11. In Paragraph 10, A terminal characterized in that the second frequency band has a starting frequency that is higher than the end frequency of the first frequency band by a preset frequency, and has a frequency bandwidth wider than the bandwidth of the first frequency band.

12. In the method using a base station, A step of receiving first radar data for a first frequency band based on a reflected signal of a signal for data transmission and sensing from a terminal; A step of generating synthetic radar data related to the first radar data through sequential processing of a first model based on VAE (Variational Autoencoder), a second model based on DM (Diffusion Model), and a third model based on GAN (Generative Adversarial Networks); and The method includes the step of transmitting second radar data including the above synthetic radar data to the terminal, A method in which the above-mentioned synthetic radar data is radar data synthesized based on the above-mentioned first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

13. In Paragraph 12, A terminal characterized in that the second frequency band has a starting frequency that is higher than the end frequency of the first frequency band by a preset frequency, and has a frequency bandwidth wider than the bandwidth of the first frequency band.

14. RF (Radio Frequency) transceiver; and It includes a processor connected to the above RF transceiver, and The processor controls the RF transceiver to receive first radar data for a first frequency band based on a reflected signal of a signal for data transmission and sensing from a terminal, generates synthetic radar data related to the first radar data through sequential processing of a first model based on VAE (Variational Autoencoder), a second model based on DM (Diffusion Model), and a third model based on GAN (Generative Adversarial Networks), and transmits second radar data including the synthetic radar data to the terminal. The above-mentioned synthetic radar data is radar data synthesized based on the above-mentioned first radar data to reflect the characteristics of a second frequency band higher than the first frequency band, and is a base station.

15. In a method performed by a terminal, A step of transmitting a signal for data transmission and sensing to a base station; A step of acquiring first radar data for a first frequency band through the reception of a reflected signal for the above signal; and The method includes the step of generating second radar data related to the first radar data through sequential processing of a first model based on VAE (Variational Autoencoder), a second model based on DM (Diffusion Model), and a third model based on GAN (Generative Adversarial Networks). A method in which the above-mentioned synthetic radar data is radar data synthesized based on the above-mentioned first radar data to reflect the characteristics of a second frequency band higher than the first frequency band.

Images